漢語右部節點提升結構

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. 漢語右部節點提升結構 On the Syntax of Right Node Raising Constructions in Mandarin Chinese. 指導教授：丁仁博士 Advisor: Dr. Jen Ting 研究生：鄭一歆 Graduate: Yi-hsin Cheng. 中華民國一零四年七月 July 2015.

(2) 中文摘要. 右部節點提升結構是跨語言的現象，有三種主要的分析法：橫越式位移、音韻刪略、多重支配。本研究旨在探討漢語右部節點提升結構，並與英語對應句做比較，進而提出分析。漢語和英語在此結構呈現以下相同的特徵：右部限制、非詞組性、並列結構限制、嚴格解/鬆散解、不對稱負極詞認可、關係修飾語認可。而兩語言在此結構上的不同之處，則是漢語表現出不對稱孤島效應以及不對稱介詞懸空。本研究發現，漢語右部節點提升結構的特徵，可由音韻刪略來得到完善解釋。. 關鍵字：右部節點提升、橫越式位移、音韻刪略、多重支配. i.

(3) ENGLISH ABSTRACT. Right Node Raising (RNR) construction is observed cross-linguistically and is generally analyzed under three main approaches, namely the across-the-board (ATB) movement approach, the PF-deletion approach, and the multidominance approach. The goal of the thesis is to examine Chinese RNR by comparing it with the English counterpart and offer an account for Chinese RNR. We observe that Chinese RNR shares some properties with the English counterpart as follows: Right Edge Restriction, non-constituency, Coordinate Structure Constraint, strict/sloppy readings, asymmetric licensing of NPI, and licensing of relational modifiers. On the other hand, Chinese RNR is different from English RNR in terms of asymmetric island effects and asymmetric P-stranding. We propose that Chinese RNR is derived by the PF-deletion approach. Under our analysis, all the properties of Chinese RNR can be accounted for.. Keywords: Right Node Raising (RNR), ATB-movement, PF-deletion, multidominance. ii.

(4) ACKNOWLEDGEMENTS. First of all, I would like to thank Professor Jen Ting for her instruction and patience in offering a wealth of enlightening and thoughtful comments. She has been responsive and instantly replied to my email with detailed information and references. I have learned a great deal from her. Sincere thanks are also due to my committee members, Professor Hsiao-hung Iris Wu and Professor Jui-heng Huang for their comments, time, and insight. There are some people I would like to thank personally. Extremely thanks are also due to my colleagues who have helped me deal with the pressure from both work and studies: Shu-ru Cheng, Ya-chi Chan, Shu-fen Pan, and Chun-fang Teng; my best friends, Kai-ting Hung, Ya-chunYu, and Wen-chi Chung, whose support and encouragement mean so much to me; Hsun Cheng, who has been my best friend since college and has accompanied me through many ups and downs; Yu-ru Shen, whose lively characteristic gives me a lot of strength. Special thanks to Mom, Dad, Pony and Danny, who are always by my side and support me whole-heartedly. In the end, from Professor Jen Ting to my committee members, and from my family to my colleagues and friends, with my deepest thanks and love.. iii.

(5) TABLE OF CONTENTS Chinese Abstract ..................................................................................................................... i English Abstract ..................................................................................................................... ii Acknowledgements ...............................................................................................................iii Table of Contents .................................................................................................................. iv List of Tables ......................................................................................................................... vi Chapter 1 Introduction .......................................................................................................... 1 Chapter 2 Properties and Previous Analyses of the RNR in General ............................... 5 2.1 General properties of English RNR ................................................................................. 6 2.2 The ATB-movement approach ..................................................................................... 19 2.2.1 Advantages .............................................................................................................. 20 2.2.2 Disadvantages .......................................................................................................... 22 2.2.3 Summary ............................................................................................................... 29 2.3 The PF-deletion approach ............................................................................................ 30 2.3.1 Advantages ............................................................................................................ 32 2.3.2 Disadvantages ........................................................................................................ 37 2.3.3 Summary.................................................................................................................. 38 2.4 The MD approach ......................................................................................................... 38 2.4.1 Advantages ............................................................................................................ 39 2.4.2 Disadvantages .......................................................................................................... 43 2.4.3 Summary ................................................................................................................ 44 2.5 The external remerge approach .................................................................................... 44 2.6 The eclectic approach .................................................................................................. 47 2.7 The sparse approach ..................................................................................................... 52 2.8 Chapter summary ........................................................................................................... 55 Chapter 3 Properties and Previous Analyses of the Chinese RNR ................................. 57 3.1 General properties of Chinese RNR ............................................................................. 57 3.1.1 Similarities between English RNR and Chinese RNR .......................................... 57 3.1.2 Differences between English RNR and Chinese RNR .................................. ……67 3.1.3 Other properties of Chinese RNR .................................................................. ……72 iv.

(6) 3.2 Two previous works on Chinese RNR ......................................................................... 79 3.2.1 Cheng (2012) ......................................................................................................... 79 3.2.2 Wang (2014a, b) .................................................................................................... 87 3.2.3 Summary ................................................................................................................ 98 3.3 Chapter summary ......................................................................................................... 99 Chapter 4 Toward a PF-deletion Approach .................................................................... 101 4.1 Against other approaches ........................................................................................... 101 4.1.1 Against the ATB-movement approach ................................................................ 101 4.1.2 Against the external remerge ............................................................................... 105 4.1.3 Against the MD approach .................................................................................... 106 4.1.4 Against the eclectic approach .............................................................................. 110 4.1.5 Against the sparse approach ................................................................................ 111 4.1.6 Summary................................................................................................................ 111 4.2 Toward a PF-deletion approach ................................................................................... 112 4.3 Implications .................................................................................................................. 126 4.4 Chapter summary ......................................................................................................... 128 Chapter 5 Conclusion .......................................................................................................... 130 References ............................................................................................................................. 131. v.

(7) LIST OF TABLES Table 1 General Properties of English RNR.................................................................................. .19 Table 2 Properties Explained by Sabbagh’s version of the ATB-movement Approach .......... 30 Table 3 Properties Explained by the PF-deletion Approach ........................................................ 38 Table 4 Properties Explained by the MD Approach ...................................................................... 44 Table 5 Properties Explained by the External Remerge Approach ............................................. 47 Table 6 Properties Explained by the Eclectic Approach ............................................................... 51 Table 7 Properties Explained by the Sparse Approach ................................................................. 55 Table 8 Summary of Different Approaches to RNR ..................................................................... 56 Table 9 Similarities and Differences between the English and Chinese RNR........................... 79 Table 10 Properties Explained by Cheng’s MD approach............................................................ 87 Table 11 Properties Claimed to be explained by Wang’s approach with TEC plus leftward focus Movement .................................................................................................................... 98 Table 12 Summary of the two works on Chinese RNR ..............................................................100 Table 13 Summary of the approaches to Chinese RNR ..............................................................126. vi.

(8) Chapter 1. Introduction This thesis investigates the nature of Mandarin Chinese Right Node Raising (henceforth RNR) construction by comparing it with the English counterpart and to propose a PF-deletion approach to Chinese RNR. RNR is first observed in Ross (1967) and is termed in Postal (1974) as a coordinate structure in which a shared element (i.e. the RNR target) in the coordinate conjuncts appears to be at the rightmost edge of the sentence and it is interpreted as in the rightmost edge of each conjunct. As (1) shows, the DP syntax is shared by both conjuncts and it occurs at the sentence-final position. The interpretation of (1) is that John loves syntax, but Bill hates syntax.. (1) John loves but Bill hates, syntax.. RNR is common cross-linguistically as in English, German, Chinese, Tagalog, Russian, Japanese, Korean, Irish, French, Polish, Spanish and so on (cf. McClosky 1986, Larson 2011a). In particular, English RNR has long been of interest to generative grammar under three main approaches, namely across-the-board (ATB) movement, PF-deletion, and multidominance (MD). Recently, an external remerge approach, an eclectic approach and a sparse approach have been proposed as well. Although RNR seems to be rich in natural languages, Chinese RNR has not been well-discussed in the literature. To the best of our knowledge, only Cheng‘s (2012) multidominance and Wang‘s (2014a, b) approach with True Empty Category (TEC) plus the leftward focus movement have been proposed. In this thesis, we show that Chinese RNR as in (2) and English RNR as in (1) share the following properties: Right Edge Restriction, non-constituency, Coordinate Structure Constraint, strict/sloppy readings, asymmetric licensing of NPI, and licensing of relational modifiers. 1.

(9) (2) Zhangsan Zhangsan. xihuan. dan Lisi taoyan. jufa. like. but Lisi hate. syntax. ‗Zhangsan likes but Lisi hates syntax.‘. Nevertheless, differences between Chinese RNR and English RNR are observed as well. Chinese RNR exhibits asymmetric island effects in the first conjunct and asymmetric P-stranding in the second conjunct as in (3) and (4) respectively.. (3) a.. Zhangsan haizai kaolu. yao-bu-yao jiaru. dan Lisi mashang. Zhangsan still. yes-not-yes. but Lisi right.away reject-Asp. Wangwu de. consider. tiyi. yinwei. Wangwu DE suggestion. join ta. because he. taoyan. de. Mali. hate. DE Mary. jujue-le. ye. jiaru-le. also join-Asp. bianlun she debate. club. ‗Zhangsan still considers whether to join or not, but Lisi rejected Wangwu‘s suggestion right away because Mary who he hates also joined the debate club.‘ b. *Lisi mashang. jujue-le. Lisi right.away de. Mali. ye. DE Mary. Wangwu de. tiyi. yinwei. reject-Asp Wangwu DE suggestion jiaru-le. because he. dan Zhangsan haizai kaolu. also join-Asp but Zhangsan still. ta. taoyan hate. yao-bu-yao. consider. jiaru. yes-not-yes join. bianlun she debate. club. ‗Lisi rejected Wangwu‘s suggestion right away because Mary who he hates also joined, but Zhangsan still considers whether to join the debate club or not.‘ (4) a.. *Zhangsan. tanlun. guanyu. er. Zhangsan. talk. about. and Lisi 2. Lisi. hulue. ni. zai bali. ignore. you in. Paris.

(10) yujian. de nansheng. meet. DE man. ‗Zhangsan talked about and Lisi ignored the man you met in Paris.‘ b.. Lisi hulue. er. Zhangsan. Lisi ignore. and Zhangsan. yujian. de nansheng. meet. DE man. tanlun. guanyu. ni. zai bali. talk. about. you in. Paris. ‗Lisi ignored and Zhangsan talked about the man you met in Paris.‘. On the other hand, English RNR shows asymmetric vehicle change where the R-expression John and pronoun him are interchangeable in the ellipsis site as in (5), scope ambiguity where the RNR target can scope over the coordination (i.e. inverse scope: every> someone) as in (6), and summative agreement effects where the plural form of the RNRed verb ‗have‘ agrees with a plural antecedent made up of the two disjointed subject DPs ‗Bill‘ and ‗John‘as in (7).. (5) a.. Hei hopes Mary won‘t, but the secretary knows that she will, fire Johni.. b.. *Hei hopes Mary won‘t fire Johni, but the secretary knows that she will fire Johni.. c.. Hei hopes Mary won‘t fire himi, but the secretary knows that she will fire Johni.. (6) John knows someone who speaks, and Bill knows someone who wants to learn, every Germanic language. (someone > every; every> someone). (Sabbagh 2007). (7) [Sue‘s proud that Bill[SG] _____ ] and [Mary‘s glad that John[SG] ____] have [PL]/?* has[SG] traveled tBill/John to Cameroon.. (Grosz 2014). Regarding the properties of Chinese RNR, Cheng (2012) cannot account for the asymmetric phenomena, namely island effects, P-stranding, and NPI licensing; Wang (2014a, 3.

(11) b) inaccurately concludes the subcategorization requirement of the RNR target and island insensitivity. We argue that asymmetric island effects could be attributed to a more general constraint on the coordinate structures: deletion within an island is unacceptable. This indicates that the RNR target is in the second conjunct. In addition, given that Chinese does not allow P-stranding in non-RNR sentences, the unavailability of P-stranding in the first conjunct of RNR shows that the target stays inside the second conjunct. With the empirical data, we argue against the ATB-movement, the MD, the external, the eclectic, and the sparse approaches. Rather, following Ha (2008a, b, c), we propose that Chinese RNR is derived by the PF-deletion approach. Importantly, our analysis captures not only asymmetric NPI licensing but also the two distinct properties of Chinese RNR, namely asymmetric island effects and asymmetric P-stranding. Furthermore, our analysis implicates that the English counterpart may need a hybrid analysis of the PF-deletion and the MD. The thesis is organized as follows. Chapter 2 first gives a description of English RNR and then reviews previous analyses based on their advantages and disadvantages. Chapter 3 provides a description of Chinese RNR data and also discusses the two previous works on Chinese RNR. Chapter 4 gives detailed arguments in favor of a PF-deletion approach—as our proposal and gives further implications on English RNR. Finally the thesis is concluded in chapter 5.. 4.

(12) Chapter 2. Properties and Pervious Analyses of the RNR in General English RNR has been widely discussed to exhibit certain properties and has received at least three main analyses. The first proposal is the ATB movement approach, arguing that the RNR target in each conjunct is moved across the board to the sentence external position, the rightmost edge of the sentence (Ross 1967, Postal 1974, Sabbagh 2007). Secondly, the PF-deletion approach claims that there are two RNR targets and the one in the first conjunct is phonologically elided under certain identity condition of the element in the second conjunct (Wexler and Culicover 1980, Kayne 1994, Hartmann 2000, Bošković 2004, Féry and Hartmann 2005, An 2007, Ha 2008a, b, c). Thirdly, the MD approach proposes that the RNR target is a single entity which is simultaneously dominated by multiple mother nodes (McCawley 1982, McCloskey 1986, Wilder 1999, Abels 2004, Bachrach and Katzir 2009, Grosz 2014). In addition to these three approaches, recently, Kluck (2007) proposes an external remerge approach; Barros and Vicente (2011) and Sabbagh (2012) argue for an eclectic analysis composed of the PF-deletion approach and the MD approach, and the RNR is viewed as a cover term for variant operations with identical outputs in surface. Last but not the least, Larson (2011a) and his subsequent works propose a sparse approach, claiming that there is no c-command relation between the RNR target and the first conjunct. Each of these approaches has their advantages in accounting for some cases of RNR; however, at the same time, they fail to explain other cases of RNR. The chapter is organized as follows. Section 2.1 presents the general properties of English RNR. The ATB-movement, PF-deletion and MD approaches are discussed in section 2.2, 2.3 and 2.4 respectively. The external remerge approach, the eclectic approach and the sparse approach are presented in section 2.5, 2.6 and 2.7. Finally section 2.8 sums up the chapter. 5.

(13) 2.1 General Properties of English RNR Right Edge Restriction The Right Edge Restriction (RER) has been observed in RNR (cf. Postal 1974, McCawley 1982, Wilder 1997, Hartmann 2000, among others). RER states that only the element at the rightmost edge within each conjunct can be the legitimate RNR target and thus forms an acceptable RNR as in (1). In (2a) and (2b), the RNR targets are not rightmost in both of the conjuncts, giving rise to ungrammaticality.. (1) a.. Ernest suspected _____, Louise believed _____, and Michael proved _____, that she was guilty.. b. (2) a. b.. (Postal 1974). John likes _____, but Mary dislikes _____, that car. *John can _____ your book and Mary will read the paper.. (Hartmann 2000) (Wilder 1999). *Joss will donate _____ to the library, and Maria will donate several old novels to the museum.. (Sabbagh 2007). Non-constituency Constituents are generally defined as those that can undergo leftward movement (i.e. wh-question) or be clefted. Abbott (1976) observes that RNR is not limited to constituents. Besides the constituents such as a DP in (3a), a PP in (3b) and a VP in (3c), RNR targets can affect non-constituents. Examples in (4) show that the RNR targets are not types of constituents available for leftward movement: (4a) and (4b) contain the sequences of objects and adverbials, (4c-4d) are even below word level, being derived in compositional morphology.. (3) a. John loves but Mary hates your best friend. b. Ben arrived and Peter left at 5 p.m. 6.

(14) c. I think that I would, and I know that Bill will buy a painting of cats. (4) a. Mary baked, and George frosted, [20 cakes][in less than an hour].. (Abbott 1976). (cf.* [Twenty cakes in less than an hour] Mary baked.) b. John borrowed and Bill stole [a large amount of money] [from the bank]. (cf. [A large amount of money from the bank] John borrowed.) c. This analysis suffers from under- and over-generation.. (Ha 2008a) (Husband 2015). (cf. *[Generation] this analysis suffers from under- and over-.) d. His theory over- and her theory under-generates.. (Sabbagh 2007). (cf. *[Generates] his theory over-.). Coordinate Structure Constraint (CSC) Ross (1967) observes that movement out of conjuncts applies simultaneously, as stated in (5). Within a coordinate structure, extraction of one or more conjuncts is not allowed.. (5) Coordinate Structure Constraint (CSC). (Ross 1967). In a coordinate structure, no conjunct may be moved, nor may any element contained in a conjunct be moved out of that conjunct.. Sentences (6a) and (6c) are coordinate structures joined with and. Whereas the movement of what out of the first conjunct in (6b) is illicit, the movement of what out of both conjuncts in (6d) is licit. This is because the former violates CSC while the latter respects it.. (6) a. John cooked lunch and washed the dishes. b. *Whati did John cook ____i and wash the dishes? c. John washed the dishes and Bill dried them. d. Whati did John wash ____i and Bill dry ____i? 7.

(15) As a coordinate structure, RNR obeys CSC (Ross 1967, Postal 1998, Sabbagh 2007). Each example in (7) involves three conjuncts, but only (7c), with a gap in each conjunct being associated with the RNR target, is well-formed.. (7) a.. *? Josh was looking for the dean‘s office, Maria was waiting in ____, and reporters were trying to find ____, Joss’ office.. b.. (Sabbagh 2007). * Josh was looking for ____, Marian was waiting in the dean‘s office, and reporters were trying to find ____, Joss’ office.. c.. (Sabbagh 2007). Josh was looking for ____, Marian was waiting in ____, and reporters were trying to ____, Joss’ office.. (Sabbagh 2007). Insensitivity to Right Roof Constraint (RRC) Rightward movement (i.e. Heavy NP Shift) in English is upward-bound, limited to the first cyclic nodes (i.e. must be within vP, CP, and PP) as shown in (8). In other words, rightward movement of an embedded clause is not allowed (Ross 1967).. (8) Right Roof Constraint (RRC) a.. (adapted from Sabbagh 2007). Rightward movement may move and right-adjoin an element X to the cyclic node in which X is merged, but no further.. b.. vP, CP and PP are cyclic nodes.. RNR disrespects the RRC. Consider the contrast between (9) and (10). A rightward movement of a DP to cross the cyclic node vP is grammatical in (9a); however, in (9b), the movement of a PP in the first conjunct to cross beyond the cyclic node PP is ungrammatical.. (9) Heavy NP Shift 8.

(16) a.. John [vP will eat ____ raw], almost anything you give him.. (Sabbagh 2007). b.. *Josh [vP elided [a review [PP of an article ___] for Jamie]], about verb-movement. (cf. John edited [a review ___] for Sue, of an article about verb-movement.) (Sabbagh 2007). Sentences in (10) demonstrate the violations of RRC (i.e. escape two CPs) in RNR but they are still grammatical. Seen in (10a), the DP the dean‟s office out of the PP is fine; in (10b), the movement of the DP out of the VP is acceptable, too.. (10) RNR a.. [CP [TP Joss [VP walked suddenly into ____]], and [TP Maria [VP stormed quickly out of ____]], the dean’s office].. b.. (Sabbagh 2012). [CP [TP Josh [VP promised [CP that he would [VP give ____ to Jamie]]]], and [TP Joss [VP claimed [CP that he was going to [VP give ____ to Sue]]]], all of the answers to the final exam].. (Sabbagh 2012). From the above, the crucial difference between Heavy NP Shift and RNR is that the former respects the RRC while the latter does not.. Island insensitivity It is well-attested that movement is sensitive to island conditions. Extraction of any element out of a wh-question, a relative clause, a complex NP, and an adjunct is not allowed and will cause ungrammaticality. However, Wexler and Culicover (1980) observe that English RNR can escape islands. See the comparison below.. (11) Relative clause island 9.

(17) a.. Mary knows [a man who buys ____ ] and Bill knows [a man who sells ____ ], pictures of Fred.. b.. (Wexler and Culicover 1980). *What does Mary know [a man who buys _____] and Bill know [a man who sells ____ ]?. (Wexler and Culicover 1980). (12) Wh-island a.. John wonders [when Bob Dylan wrote _____ ], and Mary wants to know [when he recorded ____], his great song about the death of Emmett Till.. b.. (Abels 2004). * What does John wonder [when Bob Dylan wrote ____ ] and Mary want to know [when he recorded _____]?. (Abels 2004). (13) Adjunct island a.. b.. Josh got angry [after he discovered____], and Willow quit [after finding out about ____] the company’s pro-discriminatroy policy.. (Abels 2004). * What did Josh get angry after he discovered?. (Abels 2004). In the (a) sentences from (11-13), the RNR target is displaced from the relative clause, wh-island, and an adjunct respectively, but the sentences are not affected by island conditions. On the other hand, in the (b) sentences from (11-13), extraction of what violates island conditions. Examples (11-13) points to the same thing: the operation in RNR is different from wh-movement.. Preposition Stranding (P-stranding) McCloskey (1986) observes that P-stranding is unacceptable except for RNR in many languages, such as Irish,1 French, Spanish, and Polish. In languages where P-stranding is 1. The Irish examples are from McCloskey (1986): (i) *Bhí mé a éisteacht le ti inné [DP clár mór fada ar an ráidió faoin Toghachá]i Was I listen(prog) with yesterday program great long on the radio about-the election ‗I was listening yesterday to a great long program on the radio about the election.‘ 10.

(18) banned, the complement of a preposition, however, can be the RNR target. In English, P-stranding is allowed except in the Heavy NP Shift construction. Consider (14) and (15).. (14) a.. Whati are they talking about ______ i?. b.. This is the towni which the old man lives in ______ i.. c.. The housei was broken into ______ i.. (15) Heavy NP Shift a.. *They talked about yesterday the election.. b.. * The old man lived in last year the town.. The prepositions can be separated from their complements as in wh-movement (14a), the relative clause (14b), and the passive (14c). In contrast, P-stranding in the Heavy NP Shift construction causes the ungrammaticality as in (15). Turn to the P-stranding in RNR. It is fine for the preposition to be stranded in either of the two conjuncts as in (16a-b) or in both conjuncts as in (17).. (16) a.. John talked about, and Mary ignored, the man you met in Paris. (Bošković 2004). b.. Mary ignored, and John talked about, the man you met in Paris. (Bošković 2004). (17) Jason walked suddenly into, and May stormed quickly out of, the dean’s office.. Thus far, two differences between the RNR and Heavy NP Shift construction are worth emphasizing: (i) the former allows P-stranding while the latter does not, and (ii) the former violates Right Roof Constraint while the latter obeys it. Note that the Heavy NP Shift (ii) Nil séin aghaidh an dlí a thuilleadh a bheith ag éisteacht le Is-not it against the law anymore be(-fin) listen(prog) with nó ag breathnu ar [DP ráidió agus teilifís an Iarthair] or look(prog) on radio and television the West(gen) ‗It is no longer against the law to listen, or to watch, Western radio and television.‘ 11.

(19) construction is generally accounted for under the rightward movement approach.. Strict/sloppy readings Sag (1976) notes that an elided pronoun in VP ellipsis allows strict and sloppy readings, a typical characteristic of ellipsis, as indicated in (18).. (18) John likes his father, but Bill doesn‘t like his father.. (Ha 2008b). a. John likes John‘s father, but Bill doesn‘t like John‘s father.. (strict). b. John likes Bill‘s father, but Bill doesn‘t like Bill‘s father.. (strict). c. John likes John‘s father, but Bill doesn‘t like Bill‘s father.. (sloppy). The pronoun his is possible to refer to the subject in the first conjunct (i.e. John) as in (18a) or the subject in the second conjunct (i.e. Bill) as in (18b), inducing the strict reading. Moreover, his can refer to the subject in each conjunct simultaneously (i.e. John and Bill), the so-called sloppy reading. Just as VP ellipsis, Kimura (1985) observes strict and sloppy ambiguity in the RNRed pronouns. Shown in (19), his can refer to either Bob or Tom as in (19a) and (19b) respectively, or both Bob and Tom as in (19c).. (19) Bobi knows, but Tomj doesn‘t know, how to crane hisi/j/(i,j) neck.. (Kimura 1985). a. Bob knows how to crane Bob‘s neck, but Tom doesn‘t know how to crane Bob‘s neck. b. Bob knows how to crane Tom‘s neck, but Tom doesn‘t know how to crane Tom‘s neck. c. Bob knows how to crane Bob‘s neck, but Tom doesn‘t know how to crane Tom‘s neck.. Besides, RNRed pronouns can be interpreted as simultaneously bound by a quantifier expression (e.g., every) in each conjunct (Jacoboson 1999). (20) Every Americani loves ___, and every Germanj hates ___, the town he(i,j) was born in. 12.

(20) (Sabbagh 2007) (21) Johni likes, but Billj hates, drawing pictures of himself(i,j).. Example (20) has the interpretation that every American loves the town he (American) was born in, and every German hates the town he (German) was born in. Similarly, a RNRed reflexive has a sloppy reading as well. In (21), himself refers to both John and Bill. Taken together, pronouns and reflexives contained in the RNR target allow a strict and a sloppy reading.. Asymmetric vehicle change First look at the example of VP ellipsis in (22). If John and he in (22a) are coreferential, the reconstruction of the VP loves John at LF would violate Principle C as in (22b). To solve the potential problem of Principle C violation, Fiengo and May (1994) propose an analysis as in (22c), where the R-expression John can be realized as the pronoun he.. (22) a. Mary loves Johni and hei thinks Sally does, too.. (Fiengo and May 1994). b.* Mary loves Johni and hei thinks Sally loves Johni.. (Fiengo and May 1994). c. Mary loves Johni and hei thinks Sally loves himi too.. (Fiengo and May 1994). According to Fiengo and May, the distinction between pronominal and non-pronominal elements is the feature [± pronoun], and the reconstruction of VP ellipsis is sensitive to identical value of this feature. R-expressions have the feature [-pronoun] which can be realized as [＋pronoun]. Therefore, (22c) can be derived by allowing John ([－pronoun]) to undergo featural vehicle change to him ([＋pronoun]). This formulation and the specific schema are given in (23) and (24) respectively. (23) Vehicle Change. (Fiengo and May 1994) 13.

(21) Nominals can be treated as non-distinct with respect to their pronominal status under ellipsis. (24) [－pronominal] = e [＋pronominal]. (Fiengo and May 1994). (where =e means ‗forms an equivalence class under ellipsis with‘). Likewise, Fiengo and May also assume that pronouns and reflexives are nondistinct in the reconstruction since both have the identical feature [＋pronoun] and they are the same argument. Consider (25).. (25) a.. Johni didn‘t speak for himselfi, but Susan did.. b.. *Johni didn‘t speak for himselfi, but Susan spoke for himselfi.. c.. Johni didn‘t speak for himselfi, but Susan spoke for himi.. Example (25a) is reconstructed as (25b) at LF, in which himself in the second conjunct is bound by John in the first conjunct, violating Principle A. Yet after vehicle change is applied at LF, himself can be realized as him. Therefore, the Principle A violation can be avoided as shown in (25c). Just as VP ellipsis, Ha (2008b) observes the cases of RNR in which Principle C violations can be rescued by vehicle change. Consider (26).. (26) a.. Hei hopes Mary won‘t, but the secretary knows that she will, fire Johni.. b.. *Hei hopes Mary won‘t fire Johni, but the secretary knows that she will fire Johni.. c.. Hei hopes Mary won‘t fire himi, but the secretary knows that she will fire Johni.. The reconstruction of (26a) is in (26b), where the RNRed R-expression John is bound by the pronoun he in the first conjunct, violating Principle C. Through vehicle change, [－pronoun] John can be realized as [＋pronoun] him and thus avoids the violation. 14.

(22) Larson (2011a) points out the asymmetry that only when a pronoun occurs in the first conjunct can it refer to the RNRed proper name.. (27) *The secretary hopes that Susan won‘t, but hei knows that she will, fire Johni. (Larson 2011a). Seen in (27), if the pronoun he is in the second conjunct, it cannot bind the RNRed proper name. The grammatical contrast between (26a) and (27) shows that the application of vehicle change is asymmetric.. Asymmetric licensing of negative polarity items (NPI) Negative polarity items (NPI) can be part of RNR targets, as any in (28), where a licensor is contained in each conjunct.. (28) a.. Mary couldn‘t sell, and John wouldn‘t buy, any books about linguistics. (Kayne 1994). b.. Nobody enjoyed, and few people even liked, any of the talks on Right Node Raising.. (Phillips 1996). Furthermore, Kayne (1994) and Hartmann (2000), among others, observe that asymmetric NPI licensing in RNR. See the contrast in (29). When a licensor is contained in the second conjunct, the sentence is good as in (29a); however, when such a licensor is contained in the first conjunct as in (29b), the sentence is bad.. (29) a. b.. Some people liked, but nobody really enjoyed, any of the talks on syntax. *Nobody really enjoyed, but some people liked, any of the talks on syntax. 15.

(23) Example (30) demonstrates the asymmetric licensing of NPI as well.. (30) a. b.. John bought, but Bill didn‘t read, any books about insects. *John didn‘t buy, but Bill read, any books about insects.. Licensing of relational modifiers Relational modifiers (e.g., same, similar, different, respectively, each other) which usually induce an internal reading (i.e. distributive reading) in the context of a plural noun or a plural predicate are observed in RNR (Gazdar 1987). However, it is not available when the relational modifiers occur in each conjunct. Consider (31).. (31) a. b.. John wrote and Bill read similar novels.. [internal. external]. John wrote similar books and Bill read similar novels.. [*internal. external]. Example (31a) has the prominent reading that the books John wrote are similar to the books Bill read. It is an internal reading in that the comparison is made between each conjunct. Additionally, (31a) has another interpretation: the books John wrote were similar to the books written by someone else, and the books Bill read were similar to the books read by someone else. This is an external reading in that the comparison is made outside each conjunct. With the RNR target occurring in each conjunct as in (31b), the sentence can only have an external reading: the books John wrote were all of a type, and the books Bill read were all of a type. In addition, Phillips (1996) notes that RNR can license reciprocals as each other in (32), referring to the subjects of both conjuncts, namely John and Mary.2. 2. Philips (1996) also notes that (32) would be ill-formed for speakers who reject split antecedents for reciprocals. 16.

(24) (32) John sold and Mary bought each other’s textbooks.. Scope ambiguity Jacobson (1999) observes that a quantifier expression in RNR induces ambiguous scope, namely the surface scope or the inverse scope. Example (33) illustrates the two readings of RNRed every.. (33) a.. John knows someone who speaks, and Bill knows someone who wants to learn, every Germanic language. (someone > every; every> someone). b.. (Sabbagh 2007). John knows someone who speaks every Germanic language, and Bill knows someone who wants to learn every Germanic language.. (someone > every) (Sabbagh 2007). c.. John knows someone who speaks every Germanic language.. (someone > every) (Sabbagh 2007). In (33a), every in each conjunct can have a surface scope, meaning that John knows a specific man who can speak every Germanic language, and Bill knows a specific woman who wants to learn every Germanic language. Besides the surface scope, it is possible for every to undergo Quantifier Raising and scopes over someone (i.e. inverse scope), meaning that every Germanic language is spoken by someone, and John knows these men; every Germanic language is wanted to be learned by someone, and Bill knows these men. On the other hand, in (33b), with the RNRed every Germanic language pronounced twice, only the surface scope is available: a man who speaks every Germanic language is known by John; the man who wants to learn every Germanic language is known by Bill. The same holds for the single clause in (33c).. 17.

(25) Summative Agreement McCawley (1982) observes the verb agreement effects in RNR as shown in (34-36).. (34) a. Historical and scientific knowledge are/*is different in nature. b. Thai and Burmese food are/*is quite similar.. (McCawley 1982) (Postal 1998). (35) The pilot claimed that the first nurse t1 and the sailor proved that the second nurse t1– *[was a spy]1/[were spies]1.. (Postal 1998). (36) [Sue‘s proud that Bill[SG] _____ ] and [Mary‘s glad that John[SG] ____] have [PL]/?* has[SG] traveled tBill/John to Cameroon.. (Grosz 2014)3. In (34), although the subject in each conjunct is singular, only the plural verb to be is licit in the sentences. McCawley thus claims that such examples are the application of a conjoined NP4 in RNR. Postal (1998) also notices the phenomenon that RNR allows verbal agreement which is accumulated, i.e. the verb agrees in plural with two singular subject DPs as in (35). Likewise, in (36), the auxiliary have is determined by the subjects in both conjuncts. The above examples are problematic and puzzling since there is no plural antecedent in each conjunct for the verb to agree with. That is, the two singular DPs do not constitute a plural antecedent at any level of representation. Husband (2015) notes that the type of coordinator has effect on the licensing of summative agreement. If RNR is a disjunction, only a single agreement is well-formed as in (37).. (37) [Sue is proud that Bill], or [Mary is happy that John], has/*have traveled to Cameroon. (Husband 2015) 3. Noted by Grosz, cumulative agreement can be found in a large number of languages, e.g., Armenian, Czech, German, Hebrew, Italian, among others. 4 But notice that Postal (1998) argues against such instances as RNR since they lack the intonational characteristics of clausal RNR cases (i.e. there is no hiatus after the conjoined adjectives). He claims that they should be regarded as conjunction reduction. 18.

(26) In this section, we have described the general properties of English RNR and summarize them in Table 1. Next, we will review different approaches to the RNR in the literature.. Table 1 General Properties of English RNR 1.. Right Edge Restriction (RER). 2.. Non-constituency. 3.. Coordinate Structure Constraint (CSC). 4.. Right Roof Constraint (RRC) violation. 5.. Island insensitivity. 6.. P-stranding. 7.. Strict/sloppy readings. 8.. Asymmetric vehicle change. 9.. Asymmetric licensing of NPI. 10. Licensing of relational modifiers 11. Scope ambiguity 12. Summative agreement. 2.2 The ATB-movement approach As assumed under the ATB-movement approach, there are two copies of the RNR target underlyingly. These two copies simultaneously move across-the-board to a position outside the coordinate structure (i.e. adjoins to CP), leaving two gaps associated with the RNR target (Ross 1967, Abbott 1976, Postal 1998, Sabbagh 2007). During the movement, the two copies are reduced to one RNR target. For instance, the RNR target syntax in (38a) is argued to be base-generated in the rightmost of each conjunct. Underlyingly, there is a gap associated with the RNR target in each conjunct as in (38b). The tree structure is shown in (38c).. (38) a. b.. John loves but Bill hates syntax. John loves ti but Bill hates ti syntaxi.. 19.

(27) c.. syntax XP. XP. syntax. syntax. In general, this approach assumes RNR as a type of ATB-movement; therefore, it can capture the CSC and the properties related to the RNR target being in the high scope over the coordination, i.e. the licensing of relational modifiers and scope ambiguity. On the other hand, it is peculiar why the RNR violates locality constraints on movement.. 2.2.1 Advantages The traditional ATB-movement approach (Ross 1967, Abbott 1976, Postal 1974) successfully explains the CSC, the licensing of relational modifiers, and scope ambiguity. The basic assumpotion of this approach is the movement of the RNR target out of each conjunct at the same time, and thus the CSC is respected. Consider (39). Examples (39a) and (39b) are ruled out while (39c) is ruled in. This is because the formers follow CSC while the latter does not.. (39) a.. *?Josh was looking for the dean‘s office, Maria was waiting in ____, and reporters were trying to find ____, Joss’s office.. b.. *Josh was looking for ____, Maria was waiting in the dean‘s office, and reproters were trying to find ____, Joss’ office.. c.. (Sabbagh 2007). (Sabbagh 2007). Josh was looking for ____, Maria was waiting in ____, and reporters were trying to 20.

(28) find ____, Joss’s office.. (Sabbagh 2007). Moreover, the internal reading (i.e. distributive reading) induced by relational modifiers is explained since the RNRed modifers can scope above the whole coordiante conjuncts.. (40) John hummed, and Mary sang, the same tune/a different tune.. (Sabbagh 2007). (41) [the same tune/a different tune] [John hummed and Mary sang].. As a type of ATB-movement, the LF structure of (40) is shown in (41), where the modifer scopes higher than the coordinate structure: regarding same—the identical tune, Johh hummed and Mary sang; regarding different—John hummed a tune which is distinct from the tune Mary sang. Likewise, scope ambiguity is predicted after overtly moving the RNR target out of coordination (i.e. scope over the coordinate conjuncts). See (42) for the illustration.. (42) Some nurse gave a flu shot to ____, and administered a blood test for _____, every patient who was admitted last night. (some > every; every > some). (Sabbagh 2007). As can be seen in (42), the subject DP can scops over the whole coordination (some > every), yielding the surface scope, in which a specific nurse gave a flu shot to and admininstered a blood test for all patients. In addition, the RNRed universal quantitifer can scope over the subject DP (every > some), obtaining the inverse scope, where a different nurse gave treatment to a different paitent. Since the ATB-movement approach assumes that the RNR target to be raised to adjoin the CP, the inverse scope of the quantifiers can be directly accounted for. Last but not the least, Sabbagh (2007) prodives the adjunct extraction from NP as an 21.

(29) important argument to support the ATB-movement analysis.. (43) I sent [a book ti] to Jamie from my private collectioni.. (Sabbagh 2007). Shown in (43), the adjunct extraction to the right is licit and is analyzed as involving covert movement of the adjunct‘s host NP (or DP) and a ‗late merger‘ of the adjunct (cf. Fox and Nissenbaum 1999, Fox 2002). That is, after the covert movement, the extracted adjunct can be properly head governed by a local head V, obeying Proper Head Government (cf. Culicover and Rochemont 1992). See (44) for the illustration of RNR involving adjunct extraction.. (44) *? Josh spoke with [a student of math ____], and I chatted with [a student of physics ___] (who was) from France.. (Sabbagh 2007). The illicit (44) is because of no available host DP for the adjunct to be late merged with. If the covert movement of DP (to an external position of the coordiantion) takes place, the extracted adjunct has two DP to be merged with, rather than one single DP. As a result, (44) is ill-formed. The contrast between (43) and (44) favors the ATB-movement approach to RNR.. 2.2.2 Disadvantages Problems arising from the tranditonal account of the ATB-movement approach are the violations on movement constraints (i.e. non-consistuency, RRC violation, island insensitivity, and P-stranding), strict/sloppy readings, asymmetric vehicle change, asymmetric licensing of NPI, and summative agreement. First of all, let us consider the problems related to movement constraints. Being a type of ATB-movement, it is puzzling why RNR exhibits properties that violate movement 22.

(30) restrictions. Departing from the traditional ATB-movement approach, Sabbagh (2007) solves these problems by arguing that (i) all rightward movements are unbounded as formulated in (45) and (ii) the new linear order of the syntactic objects cannot disrupt the already built ones as in (46) (cf. Fox and Pesetsky 2005).. (45) Rightward Crossing Constraint (RCC) Rightward movement of X may not cross phonologically overt material which is not contained within the cyclic node (=vP, CP) wherein X is initially merged. (46) Order Preservation The linear ordering of syntactic units is affected by Merge and Move within a Spell-Out Domain, but is fixed once and for all at the end of each Spell-Out Domain.5. According to Sabbagh, RER is the outcome of Order Preservation. When a RNR target is originated at the right edge of each conjunct and is the complement to a head (i.e. vP, PP, and CP), it can undergo ATB-movement without crashing the built linear ordering. Importantly, only an RNR target which is the complement of a head can be spelled out and thus forms a licit RNR (since the RNR target originates at the cyclic node which immediately dominates it). Non-constituency effects are illustrated in (47), where the target involves Theme and Instrument in (47a) and a sub-word in (47b).. (47) a. John has sliced, and Mary also seems to have sliced, [a large piece of cake with a shining new knife].. 5. (Sabbagh 2007). Adopting the Cyclic Spell-Out model of grammar (Bresnan 1971, Chomsky 2000, Epstein and Seely 2002), Sabbagh defines Spell-Out as (i) the operation which takes cyclic nodes (i.e. vP, PP and CP) over to the PF components, forming the ordering, and (ii) the application to the complement of the head of the Spell-Out domain (SOD). Note that once a cyclic node is spelled out, its linear order is fixed. 23.

(31) b. It is the quickest way to rescue [over-____] and [under-_____] exposure.. To account for (47a), Sabbagh adopts Larson‘s (1990) analysis of VP-structure in which the bracketed sequence is generated as a vP as in (48).. (48) [vP a large piece of cake [V‘ V with a shining new knife]]. (Sabbagh 2007). As for (47b), the raised exposure from the sub-word level Spell-Out Domain (SOD) in the first conjunct follows Order Preservation, deriving the linear order over- > exposure. The same applies to the second conjunct and leads to the correct ordering under- > exposure. As a result, the non-constituent RNR target in (47b) can be predicted as long as the movement of it does not violate Order Preservation. Turn to the RRC violation. Sabbagh claims that a RNR is still well-formed if the RNR target violating the RRC does not cross any overt element where the target originates, i.e. regardless of the RRC violation. See the contrast below.. (49) Josh said that he thought that I should sell ___, and Jamie said that she thought that she might want to buy ___, each of the Rambaldi artifacts that I have in my attic. (Sabbagh 2007) (50) *Jamie [read one long review [of a 10 page article___] and two long reviews [of a 20 page article ____] for the same journal, about my new book.. (Sabbagh 2007). Sentence (49) has the gap deeply embedded within at least two clauses in each conjunct, and it is still grammatical under the RCC. On the other hand, (50) is ill-formed because the RNR target in the second conjunct crosses not only the first cyclic node PP (where the target is originally merged) but also another cyclic node PP for the same journal. Crucially, this PP is 24.

(32) an overt phonological element (wherein the RNR target is not initially generated), and the crossing of it would inevitably violate the RCC. Island insensitivity can be captured since Order Preservation is not violated (i.e. do not cross any phonological overt elements). Consider (51).. (51) John [vP knows a girl who likes ____], and Bill [vP knows a girl who hates ____], paintings of Fred.. In (51), the RNR target paintings of Fred is originally contained in the cyclic node vP in the first conjunct, deriving the ordering who > likes > paintings of Fred. When the RNR target is extracted to the external position, the built ordering maintains. Likewise, the extraction of RNR target in the second conjunct of (51) does not affect the built ordering. Also, RCC is not violated in (51) since no overt phonological element has been crossed in each conjunct. Under Sabbagh‘s account, islands can be disregarded if the PF order is respected in RNR.6 P-stranding is predicted by following the Order Preservation. Consider (52) and its derivation in (53).. (52) Josh walked slowly into, and Jamie walked quickly out of, the dean’s office. (Sabbagh 2007) (53) a.. First Conjunct. i.. [vP1 v [vP walked [PP into DPi] slowly]]]. (Move-PP). ii.. [vP1 v [vP walked t PP slowly] [PP into DPi]]. 6. However, Bachrach and Katzir (2009) point out that RNR can feed wh-movement is a counterexample to Sabbagh‘s RCC. In (i), wh-movement is applied to the RNR target after the linearization while in ATB movement (ii), island effects are observed. To account for the contrast, Bachrach and Katzir propose a condition on ATB: wh-movement is possible across-the-board only if no island is observed in either conjunct, or if a counterpart RNR sentence is well-formed. (i) Which booki did [John meet the man who wrote ti], and [Mary meet the woman who published ti]? (ii) Whoi did [a man who loves ti dance], and [ a woman who hates ti go home]? 25.

(33) Spell-Out (vP1): walked > slowly iii. [CP1 Josh T0 [vP1 v [VP1 walked t PP slowly] [PP into DPi]]] Spell-Out (CP1): Josh > VP1 > into > DPi. (Sabbagh 2007). b.. Second Conjunct. i.. [vP2 v [vP walked [PP out-of DPi] quickly]]. ii.. [vP2 v [VP2 walked t PP quickly] [PP out-of DPi]]. (Move-PP). Spell-Out (vP1): walked > quickly iii. [Conj, and [CP Jamie T0 [vP2 v [VP2 walked t PP quickly] [PP out-of DPi]] Spell-Out (CP2): Jamie> VP2 > out-of > DPi (54). (Sabbagh 2007). New ordering: CP1> Conj‘ ; Conj > CP2 ConjP[CP] > DPi. (Sabbagh 2007). In (53a), first we apply rightward movement7 of PP ((i)). Since vP1 is a spell-out domain (SOD), we get the ordering walked > slowly ((ii)). The next SOD is CP1, rendering Josh > VP1 > into > DPi ((iii)). The same applies to the derivation of the second conjunct as in (53b). Next, two conjuncts merge together, forming the coordinate structure, ConjP as in (54). Since the ConjP is composed of coordinate CP‘s, it is an acceptable landing site for rightward movement. That is, rightward movement of DPi can apply to the rightmost edge position of the matrix CP. Finally, the linear ordering is established. Thus far, we have seen that Sabbagh‘s analysis successfully captures most of the RNR properties, especially those violating movement restrictions. Still, there are problems which cannot be solved even under his account, namely strict/sloppy readings, asymmetric vehicle 7. Sabbagh (2007) formulates the legitimate landing sites for rightward movement as (i): (i) Landing sites for rightward movement Rightward movement may move an element X: (i) to the right edge of the first Spell-Out domain that contains X‘s base position, or (ii) to the right edge of the matrix CP. 26.

(34) change, asymmetric licensing of NPI, and summative agreement. Consider sloppy reading. It is assumed that only the RNR target and the subject are within the same conjunct can we get the sloppy reading. However, under the ATB-movement approach, on the (surface) structure, the raised RNR target is no longer in the same conjunct as the coindexed subject. Therefore, the sentence is predicted to be ungrammatical, contrary to fact. Consider (55).. (55) Every Americani loves ___, and every Germanj hates ___, the town he(i,j) was born in. (Sabbagh 2007) In (55), the RNRed he is a bound variable which can be bound by the subject of each conjunct. However, the raised he and its coindexed subject are not in the same conjunct anymore, so the bound interpretation should be impossible, again, contra fact.8 Asymmetric vehicle change is puzzling under this approach. Being outside of the coordination, the R-expression John in (56) is expected not to refer to the pronoun he. Again, the ATB-movement approach fails.. (56) Hei hopes that Susan won‘t, but the secretary knows that she will, fire Johni at the end of this year.. (Ha 2008b). The licensing of NPI weakens the ATB-movement approach, let alone the asymmetric one. According to Linegarger‘s (1980, 1987) Immediate Scope Constraint, the licensing relation of NPIs and negation is sensitive to locality conditions. That is, an NPI can only be. 8. Sabbagh (2007) does not think such an argument is conclusive and further provides the counterexample on wh-movement. In (i), the pronoun his contained in the wh-phrase allows both a strict and a sloppy reading, suggesting the same phenomenon holds for wh-movement which depends on the ATB-movement analysis (Jacobson 1999, Nissenbaum 2000, among others). (i) Which of his(i,j) parentsk does every Americani love tk best, and every Germanj love tk least? (Sabbagh 2007) 27.

(35) licensed if it is in the immediate scope of negation.. (57) John didn‘t sell, and Bill didn‘t buy, any apples. (58) John read, but he hasn‘t understood any of her books.. Seen in (57), the surface position of the NPI is not in the immediate scope of the negation, so the RNR should be illicit, contra fact. Likewise, the prediction of the ill-formed (58) is incorrect. This wrong prediction under the ATB-movement analysis is left open in Sabbagh (2007) with a suggestion that “the possibility of reconstruction for RNR cannot be precluded on general grounds.” He provides the well-formed reconstructed (59) (Linebarger 1980, 1987). Nevertheless, (59) involves A-movement while RNR involves A‘-movement under Sabbagh‘s analysis. The A‘-movement of an NPI still cannot be reconstructed as in (60), showing that Sabbagh‘s example still cannot account for asymmetric NPI licensing.. (59) [A doctor with any reputation]i is likely not to be ti available.. (Sabbagh 2007). (60) *Any meat Bill didn‘t eat.. Consider (61) for the illustration of summative agreement effects.. (61) a.. [Mary is proud John ______ ] and [Sue is glad that Bill _____ ] have traveled tJohn/Bill to Cameroon.. b.. (Grosz 2014). *[Mary is proud that John T have traveled tJohn to Cameroon] and [Sue is glad that Bill T have traveled tBill to Cameroon.] ATB-movement [[[Mary is proud John tXP] and [Sue is glad that Bill tXP]] [T have traveled x to Cameroon.]]. (Grosz 2014) 28.

(36) In (61a), the plural verb have is licensed by the disjoined subjects John and Bill in each conjunct respectively. Grosz (2014) points out that if the ATB-movement approach is on the right track, (61a) should be ill-formed since the embedded subject of each conjunct is singular and only the singular verb can be licensed. This prediction is not borne out. If the sentence undergoes reconstruction, it is ill-formed as in (61b). Besides, if x is a plural trace linked to John and Bill, it is expected to be licensed in collective predicates (e.g., meet). Therefore, (62) is predicted to be good, contra fact.. (62) *[[Mary is proud John tXP] and [Sue is glad that Bill tXP] [XP T met x at the party.]] (Grosz 2014). The contrast between (62) and (63) indicates the crucial problems of the ATB-movement approach: summative agreement and anticollectivity (i.e. predicates which need a plural antecedent cannot occur in the RNR target) lead to conflicting requirements on the properties of x, the bound variable with plural number feature (Grosz 2014).. 2.2.3 Summary As shown in Table 2, Sabbagh‘s (2007) ATB-movement approach accounts for the RER, non-constituency, CSC, RRC violation, island insensitivity, P-stranding, licensing of relational modifiers, and scope ambiguity. Although his account explains the most properties of RNR in comparison with other versions of the ATB-movement approach, still, it fails to expect the RNR cases of strict/sloppy readings, asymmetric vehicle change, asymmetric licensing of NPI, and summative agreement effects. Next, we discuss the advantages and disadvantages of the PF-deletion approach.. 29.

(37) Table 2 Properties explained by Sabbagh’s version of the ATB-movement approach 1.. Right Edge Restriction (RER). . 2.. Non-constituency. . 3.. Coordinate Structure Constraint (CSC). . 4.. Right Roof Constraint (RRC) violation. . 5.. Island insensitivity. . 6.. P-stranding. . 7.. Strict/sloppy readings. ?. 8.. Asymmetric vehicle change. ?. 9.. Asymmetric licensing of NPI. ?. 10. Licensing of relational modifiers. . 11. Scope ambiguity. . 12. Summative agreement. ?. 2.3 The PF-deletion approach Under the PF-deletion approach, non-movement is argued for solving the puzzles of RNR which would arise under the traditional ATB-movement approach. It assumes that the two copies of the identical RNR target are base-generated within each conjunct. Under the syntactic and semantic licensing condition, the one in the first conjunct is phonologically elided (Wexler and Culicover 1980, Hartmann 2000, Bošković 2004, Féry and Hartmann 2005, An 2007, Ha 2008a, b, c).. (64) a. b.. John loves but Bill hates syntax. John loves syntax but Bill hates syntax.. For example, (64a) is analyzed as (64b): the RNR target syntax is elided in the first conjunct at the level of PF while overtly appearing at the rightmost of the second conjunct. The most crucial distinction between the ATB-movement approach and the PF-deletion approach is that the latter does not dislocate the RNR target from their base-generated position (Bošković 30.

(38) 2004). That is, the RNR target is in the conjunct internal position, not in the conjunct external position. We review two works under the PF-deletion approach, namely Hartmann (2000 and her subsequent work with Féry 2005) and Ha (2008a, b, c), and focus on their licensing conditions for RNR. Basically, Hartmann proposes that (i) syntactically, the conjuncts must be in an identical structure, (ii) the pre-RNR elements in the first and second conjuncts must be contrastively focused, creating a set of alternatives that are identical for both conjuncts semantically,9 and (iii) phonologically, the deletion takes place right after the contrastively focused pre-RNR elements in the first conjunct. See (65).. (65) Hans [SCHȀLT]F den Kȕrbus und Maria [KOCHT]F den Kȕrbis. Hans peels. the squash and Maria cooks. (Hartmann 2000). the squash. ‗Hans peels and Maria cooks the squash.‘. Exemplified in (65), both conjuncts involve a transitive structure, i.e. being syntactically parallel. The pre-RNR elements SCHȀLT and KOCHT show a contrastive focus. At LF, the two conjuncts mutually entail each other since they create the same set of alternatives. Finally, den Kȕrbus ‗the squash‘in the first conjunct is elided to satisfy the prosodic alignment condition. Nevertheless, the problems of Hartmann‘s proposal are pointed out in Ha (2008b) as follows. First, the identical syntactic structure wrongly rules out vehicle change effects. Secondly, it is unclear how to prohibit the RNR target in the second conjunct to be deleted under the prosodic alignment condition. As argued, the nuclear accent received on the RNR target has to be aligned at the rightmost in each Intonational Phrase, meaning that both RNR targets would be deleted. However, only the deletion of the RNR target in the first conjunct. 9. The contrastive focus would be absent if these two parts are identical, too similar or not of a same semantic kind. 31.

(39) forms the structure of RNR. Alternatively, Ha (2008a, b, c) revises Hartmann and argues for syntactic, phonological, and semantic licensing conditions in RNR. Adopting Merchant‘s (2001) ellipsis feature of sluicing (ES), ERNR feature has to (i) enter the derivation with contrastively focused pre-RNR elements in the first conjunct, (ii) instruct its sister to be unpronounced, and (iii) meet the requirement that the RNR target (in the first conjunct) and its antecedent (in the second conjunct) are mutually entailed at LF. Also note for the ERNR feature to be licensed, it has to check its feature with the closest head of C.. (66) [CP F [&P JOHN MADE[ERNR] <the spaghetti>, and Bill ATE the spaghetti.]]. As illustrated in (66), JOHN MADE are the contrastively focused elements bearing ERNR. Outside the coordination, the feature F triggers agreement relation with its closest ERNR feature. Next, the ERNR feature tells the PF not to spell out its complement the spaghetti; semantically, the RNR (clause) John made the spaghetti turns to variables at LF and is existentially closed by F-closure, yielding ƎxƎR [x R-ed the spaghetti] ; the same applies to the antecedent (clause) Bill ate the spaghetti, resulting in ƎxƎR [x R-ed the spaghetti].10 Taken together, the F-closure of the antecedent entails the RNR, and that of the RNR entails the antecedent, meaning that the two conjuncts mutually entail each other, satisfying the e-GIVENESS condition.. 2.3.1 Advantages This approach assumes that RNR involves a PF-process and no overt movement takes place; therefore, it directly solves some problems of the traditional ATB-movement approach: 10. Adopting Merchant (2001), the LF condition of mutual entailment is called e-GIVEN. 32.

(40) those related to movement constraints (i.e. non-constituency, CSC, Right Roof Constraint violation, island insensitivity and P-stranding). In particular, Ha (2008b) accounts for sloppy reading, asymmetric vehicle change, and asymmetric NPI licensing. Also note that the RER is captured independently from the licensing requirements of ERNR. That is, even if an ERNR meets the syntactic, phonological, and semantic requirements, it would be ill-formed if the RER is not obeyed. Regarding the apparent non-constituency, Ha argues that it is the outcome of the deleted elements as a whole, and RNR is in fact targets the constituents. More specifically, Ha allows multiple ERNR features and each of them instructs the PF to unpronounce its sisters, i.e. the constituents. Even though the elided RNR target as a whole does not seem to be a constituent, each deletion at PF is indeed a constituent. He uses a German example to illustrate this notion because there seems to be no direct account for such non-constituency.. (67) a. [CP Ich habe einen I. have a. Mann, [CP der DREI <Katzen besitzt>, <gekannt>],. und. man. and. who three. [CP Sie hat eine Frau, [CP die VIER Katzen she has a. woman. who four cats. cats owns. knows. besitzt], gekannt.] owns. knows. ‗I have known a man who owns three cats, and she has known a woman who owns four cats.‘. (German, Wilder 1997). 33.

(41) b.. (Ha 2008b). In (67a), the RNR target Katzen besitzt gekannt is a non-constituent sequence of NP katzen „cats‟, embedded verb besitzt ‗owns‘ and the main verb gekannt ‗knows‘. Each of them is contrastively focused and bears the ERNR feature. The D ERNR feature projects and deletes its sister, vP gekannt ‗knows‘; similarly, the NP ERNR feature projects and licenses the deletion of its sister Katzen besitzt „cats owns‟. Note that the RNR target contains focused three, which cannot be elided.The AP[ERNR] is an adjunct and cannot project further, so it only deletes its sister NP Katzen „cats‟. Finally, the sequence of Katzen besitzt gekannt is elided. Taken together, Ha claims that since each ERNR feature only deletes its sister, the elided elements are indeed a consitituent. It is worth mentioning here that the most crucial difference between Hartmann (2000 and her subsequent work) and Ha (2008a, b, c) is that regarding the RNR target, the former assumes a syntactic identity whereas the latter assumes a semantic identity which can nicely solve the problematic sloppy reading, asymmetric vehicle change, and asymmetric NPI licensing. Sloppy reading can be captured by assuming the RNRed pronoun is a bound variable. Recall that the RNRed pronoun is assumed to be a bound variable and the semantic licensing condition allows different reference for pronominals. 34.

(42) (68) JOHNi LIKES[ERNR]<hisi father>, but BILLj HATES hisj father.. (Ha 2008b). In (68), at LF, the antecedent turns to bound variables and is existentially closed as Bill λx [ x hates x‟s father] and the RNR clause as John λy [ y likes y‟s father]. The outcome of F-closure of the antecedent and the RNR clause is still the same, ƎxƎR [x R x‟s father], satisfying the semantic condition. As a result, sloppy reading is captured. Asymmetric vehicle change effects can be captured as well. Ha assumes the semantic identity is relatively flexible than the syntactic identity; therefore, the co-reference between the R-expression and the pronoun is available as long as they have the same interpretation.. (69) JOHNi hopes that Susanj WON‘T <fire himi (at the end of this year)>, but THE SECRETARY knows that shej WILL fire Johni at the end of this year.. In (69), the semantic licensing condition can be met when the deleted RNR target and the antecedent mutually entail each, meaning that him refers to John. Another type of vehicle change, the asymmetric NPI licensing, can be accounted for in a similar way. Ha assumes that an NPI and a positive polarity item (i.e. PPI) share a binary feature [± polarity indefinite]: the former has the feature [+polarity] and the latter [-polarity]. Since vehicle change is insensitive to the feature value, an NPI can be treated as a PPI under reconstruction. Consider (70).. (70) a. John read, but he hasn‘t understood any of my books.. (Ha 2008b). b.. *John read any of my books, but he hasn‘t understood any of my books.. c.. John read some of my books, but he hasn‘t understood any of my books.. 35.

(43) Example (70a) has the reconstructed representation in (70b), whose ungrammaticality can be avoided by applying vehicle change. In (70c), any is realized as some in the first conjunct (i.e. the ellipsis site), giving rise to grammaticality. Additionally, Ha claims that the licensing of relational modifiers and scope ambiguity can be accounted for. After the derivation of RNR (i.e. only one target on the surface), the single instance of the relational modifier in the second conjunct can undergo covert movement out of the coordinate structure, and thus scopes over the whole coordination at LF.11 Note that this single entity of the RNR target in the second conjunct induces an internal reading, different from the cases where the two entities of the RNR target yield an external reading.. (71) a. Peter sings and Mary whistles a similar tune. b. [A similar tune] [Peter sings and Mary whistles]. (Ha 2008b) (Ha 2008b). Seen in (71), syntactically, a similar tune can covertly move out of the coordination, where it has an inverse scope, i.e. wide scope. Therefore, the internal reading can be obtained: the tune Peter sings and the tune Mary whistles is similar. Just as the cases of relational modifiers, quantifiers can exhibit inverse scope by undergoing covert movement at LF after the derivation of RNR. Specifically, after the deletion of the RNR target in the first conjunct, the RNRed quantifier phrase in the second conjunct is available for covert movement (i.e. Quantifier Raising) out of coordination at LF, inducing an inverse scope.. (72) Every philosopher read and every linguist reviewed some paper.. 11. (Ha 2008b). Ha notes that if the relational modifiers occur overtly in the first conjunct, covert movement of the target in the second conjunct is disallowed. This is because the CSC would be violated. 36.

(44) (every >some; some > every) (73) [TP Some paper [&P [TP every philosopher read <some paper>] and [TP t‘ every linguist reviewed t]. (Ha 2008b). As seen in (72), some paper in the first conjunct is elided after valued with the head of C, deriving the only one instance of some paper in the second conjunct. Next, the target is assumed to undergo Quantifier Raising at LF, making the inverse scope (i.e. some > every) possible as in (73).. 2.3.2 Disadvantages Summative agreement effects are problematic under the PF-deletion approach. With a copy of the RNR target contained in each conjunct, the approach would wrongly predict (74a) to be ill-formed just as its undeleted counterpart in (74b), contra fact. It is unclear how the plural verb have can be licensed even though the embedded subject in each conjunct is singular.. (74) a.. [Mary is proud John ______ ] and [Sue is glad that Bill _____ ] have traveled tJohn/Bill to Cameroon.. (Grosz 2014). b.* [Mary is proud that John have traveled tJohn to Cameroon ] and [Sue is glad that Bill have traveled tBill to Cameroon.]. (Grosz 2014). In addition, Kluck and Zwart (2009) points out that Ha (2008a, b, c) fails regarding the cases of more than two conjuncts. This is because Ha claims that only the closest and uninterpretable ERNR need to check its feature with the head of C during syntactic derivation. In other words, if there are three conjuncts, only the target in the first conjunct can have its value checked and then be deleted, leaving the target in the second and the third conjunct to 37.

(45) occur on the surface structure.. 2.3.3 Summary The PF-deletion approach directly accounts for the puzzles which would arise under the traditional ATB-movement approach, namely violations of movement conditions, sloppy reading, asymmetric vehicle change, and asymmetric NPI licensing. Moreover, the licensing of relational modifiers and the scope ambiguity are captured by assuming the covert movement of the single RNR target at LF. However, it is not clear how the deletion approach captures the summative agreement effects.. Table 3 Properties explained by the PF-deletion approach 1.. Right Edge Restriction (RER). . 2.. Non-constituency. . 3.. Coordinate Structure Constraint (CSC). . 4.. Right Roof Constraint (RRC) violation. . 5.. Island insensitivity. . 6.. P-stranding. . 7.. Strict/sloppy readings. . 8.. Asymmetric vehicle change. . 9.. Asymmetric licensing of NPI. . 10. Licensing of relational modifiers. . 11. Scope ambiguity. . 12. Summative agreement. ?. 2.4 The MD approach Giving up the assumption of a single mother node, the MD approach assumes that a node can have more than one mother node. The notion of constituent sharing (i.e. the RNR target is shared by both conjuncts) is basically multiple dominance (McCawley 1982, Wilder 1999, 2008, Abels 2004, Citko 2005, 2008, 2011, Bachrach and Katzir 2009, de Vries 2013). 38.

(46) Under this approach, the single entity of RNR target is simultaneously dominated by more than one mother node. Seen in (75a), two mother nodes loves and hates simultaneously dominate the RNR target syntax as in (75b).. (75) a.. John loves but Bill hates syntax.. b.. (Wilder 1999). The crucial difference between the PF-deletion approach and the MD approach is the numbers of RNR target: the former assumes two separate copies of RNR target whereas the latter assumes only one RNR target. Under PF-deletion, something can be done to the undeleted RNR target in one of the conjuncts without influencing the other RNR target. As for the MD, since there is just one RNR target, something done to the RNR target in one position is sure to impact the RNR target in another position (Wilder 1999, Larson 2011a). That is, any operation to the RNR target will carry over to every position in the structure where it is linked to.. 2.4.1 Advantages Generally speaking, the MD approach is like the PF-deletion approach since both assume no movement involved in RNR. Therefore, movement constraints (i.e. non-constituency, CSC, RRC violation, island effects, and P-stranding) can be accounted for. Importantly, the real advantages of the MD approach over the PF-deletion approach lie in the correct prediction of summative agreement effects. Besdies, under the MD approach, the 39.

(47) licensing of relational modifiers and scope ambiguity can be captured more naturally without any additional assumptions as the PF-deletion approach. The RER effects can be explained by Wilder‘s (1999) modified version of Linear Correspondence Axiom (LCA):12 only fully-dominated elements (i.e. non-shared elements) can be linearized, and they are required to be linearized before the shared elements. Wilder‘s proposal accounts for why the RNR target can only be linearized in the rightmost of the coordination as in (76).13. (76) a. b.. [TP1 John has bought] and [TP2 Mary will read] the paper.. (Wilder 1999). * John can _____ your book and Mary will read the paper.. (Wilder1999). Now let us consider the crucial advantages of the MD approach. The syntactic or semantic effects which involve a subconstituent of the RNR target seem to affect both conjuncts simultaneously. The MD approach accounts for the licensing of relational modifiers and summative agreement effects, which the ATB-movement approach cannot explain, and which require special assumptions under Ha‘s (2008a, b, c) ellipsis account. According to Abels (2004), there is a distributive/cumulative operator (OP) which must scope over a plural predicate. Sentence (77) has the structure illustrated in (78): the operator scopes over the coordinate structure and different scopes over the operator, deriving the distributive interpretation of the sentence, i.e. internal reading.. (77) John sang, and Mary recorded, two quite different songs.. (Abels 2004). 12. Kayne‘s (1994) LCA: Let X, Y be nonterminals and x, y terminals such that X dominates x and Y dominates y. If X asymmetrically c-commands Y, x precedes y. 13 Wilder assumes the RNR target can be non-rightmost at the final conjunct as in (i) (i) John should fetch _____ and give the book to Mary. But it is unsure whether (i) is an example of RNR (i.e. the book is the RNR target) or simply a coordination of two verbs. Fox and Pesetsky (2007) also note that Wilder‘s data are far from being clear in this respect. 40.