Negative Preposing: Intervention and Parametric Variation in Complement Clauses

—11— ATLANTIS Journal of the Spanish Association of Anglo-American Studies 40.1 (June 2018): 11-37 issn 0210-6124 | e-issn 1989-6840 doi: http://doi.org/10.28914/Atlantis-2018-40.1.01 Negative Preposing: Intervention and Parametric Variation in Complement Clauses Ángel Luis Jiménez-Fernández Universidad de Sevilla [email protected] This work deals with root transformations (RTs) such as negative preposing in English and Spanish. I claim that RTs may in principle be compatible with all types of embedded clauses, regardless of whether the selecting predicate is factive/non-asserted or non-factive/asserted. Languages differ in how freely they allow RTs in various types of complements. Adopting an intervention account, according to which the movement of an operator to Spec-CP intervenes with other types of movement, including RTs, I account for the variation in the distribution of English/Spanish negative preposing by certain options made possible as a result of feature inheritance of discourse features. It is well known that the distribution of RTs in English is extremely limited, while in Spanish the same operations are possible in many more constructions. In Spanish, discourse features may be inherited from complementizer (C) to tense (T) such that negative preposing targets Spec-TP, and hence there is no intervention effect. In contrast, discourse features stay at C in English, meaning that negative preposing competes for the target position with the operator movement to CP, and this gives rise to intervention. This hypothesis is explored and validated through an experiment with informants of the two languages. Keywords: negative preposing; intervention; factivity/assertedness; feature inheritance; root transformations . . . Anteposición negativa: intervención y variación paramétrica en oraciones completivas Este trabajo trata sobre transformaciones matrices (TTMM) como la anteposición negativa en inglés y español. Defiendo que las TTMM pueden ser en principio compatibles con todo 12 ÁNGEL LUIS JIMÉNEZ-FERNÁNDEZ ATLANTIS. Journal of the Spanish Association of Anglo-American Studies. 40.1 (June 2018): 11-37 • issn 0210-6124 | e-issn 1989-6840 tipo de oraciones subordinadas, independientemente de si el predicado que las seleccione sea factivo/no asertivo o no-factivo/asertivo. Las lenguas difieren en cuanto a la flexibilidad que muestran a la hora de permitir estas TTMM en los distintos tipos de complementos. Adoptando un análisis basado en la intervención, según el cual un operador que se mueve al especificador del sintagma del complementante interviene y bloquea otros movimientos, como las TTMM, explico la variación en la distribución de la anteposición negativa en español e inglés mediante las opciones que nos ofrece la herencia de rasgos discursivos. Es bien sabido que las TTMM en inglés están muy limitadas en su distribución, mientras que en español las mismas pueden aparecer en muchas más construcciones que en inglés. En español los rasgos discursivos pueden ser heredados por la categoría tiempo (T) desde el complementante (C), de manera que la anteposición negativa conlleve el movimiento al especificador de T, sin que haya ningún tipo de intervención. Por el contrario, estos rasgos discursivos permanecen en C en inglés, y la anteposición negativa compite con el operador por la misma posición sintáctica, dando lugar a efectos de intervención. La validez de esta hipótesis se explora mediante un experimento con informantes nativos de ambas lenguas. Palabras clave: anteposición negativa; intervención; factividad/asertividad; herencia de rasgos; transformaciones matrices 13 NEGATIVE PREPOSING IN COMPLEMENT CLAUSES ATLANTIS. Journal of the Spanish Association of Anglo-American Studies. 40.1 (June 2018): 11-37 • issn 0210-6124 | e-issn 1989-6840 1. Introduction Since the publication of the seminal paper by Joseph Emonds (1969), a number of proposals have been put forth to attempt to explain why certain types of transformations can only occur in root contexts—see, among others, Hooper and Thompson (1973), Emonds (1976), Maki, Kaiser and Ochi (1999), Haegeman (2000; 2002; 2006a; 2006b; 2007 and 2010), Heycock (2006), Bianchi and Frascarelli (2010), Miyagawa (2010), Jiménez-Fernández and Miyagawa (2014). More precisely, the generalization that all these linguists draw is that certain transformations are restricted to main clauses and subordinate clauses with root properties.1 Concentrating on negative preposing (NPr), which has been described as a subtype of focus fronting in English by Andrew Radford (2009) and Haegeman (2012, 44), the following examples show that its distribution is restricted to main or root clauses (or root-like clauses) in a language such as English:2 (1) Seldom have the children had so much fun. (2) I exclaimed that never in my life had I seen such a crowd. (3) *It’s likely that seldom did he drive that car. (Hooper and Thompson 1973, 479) Note that the fronted constituents are all adjuncts, which, according to Haegeman (2012, 73), makes fronting easier. However, arguments can also be hosted in the left periphery as a consequence of negative preposing, and therefore a distinction between root and non-root contexts is also relevant: (4) Not a single book did he buy. (Haegeman 2012, 9) (5) I swear that not a single book did he buy. (Haegeman 2012, 9) (6) *It is unlikely that not a single book did he buy in all his life. In his original study, Emonds claims that “a root will mean either the highest S in a tree, an S immediately dominated by the highest S or the reported S in indirect discourse” (1969, 6). In a later work, Emonds identifies a series of transformations that can be applied in embedded contexts such as topicalization and negative preposing 1 Earlier versions of this paper have been presented at the 2017 annual meeting of the Linguistic Association of Great Britain (LAGB) (University of Kent, 4-7 September, 2017), at the thirty-first Going Romance conference (University of Bucharest, 7-9 December, 2017) and at a seminar in the University of the Basque Country (27 June, 2017). I thank the audiences there for their fruitful comments. In particular, I am grateful to Mara Frascarelli, Liliane Haegeman and Myriam Uribe-Etxebarria for our constant discussion on the issues treated here. I am also thankful to Alejo Alcaraz, Adam Ledgeway, Michelle Sheehan and Vidal Valmala, Javier Ormazabal for useful comments and suggestions on the proposal and the data presented here. The research in this paper has been funded by the Spanish Ministry of Economy and Competitiveness (project FFI2013-41509-P). Finally, I am grateful to the two anonymous reviewers of Atlantis for their insightful comments which definitely have improved the work. 2 Unless otherwise indicated, the examples are my own creation. 14 ÁNGEL LUIS JIMÉNEZ-FERNÁNDEZ ATLANTIS. Journal of the Spanish Association of Anglo-American Studies. 40.1 (June 2018): 11-37 • issn 0210-6124 | e-issn 1989-6840 in English (2004). As he remarks, these embedded root phenomena, according to Hooper and Thompson (1973), correspond to those transformations which can be applied in indirect discourse. These root-like indirect discourse embeddings—RIDEs, in Emonds’s terminology—freely allow root transformations (RTs) such as negative preposing, whereas other types of embeddings block this kind of syntactic operation. Examples (7), (8) and (9) illustrate three instances of RIDE, all taken from Emonds (2004, 77). (7) Bill warned us that [RIDE flights to Chicago we should try to avoid]. (8) John said that [RIDE never did the children help his mother]. (9) It is shocking to Sue that [RIDE not once has Mary heard from her children]. For Emonds, RIDEs are finite complement clauses of a governing verb (V) or adjective (A) (2004). This explains the free application of NPr in (7), (8) and (9). The same syntactic context allows topic fronting and RTs in general. As Emonds observes, if these operations apply in non-RIDE environments, the outcome is ill-formed. This is seen in the following examples, also from Emonds (2004, 77). (10) *We will propose [only until five working] to the management. (11) *I ignored the boss who was so angry that [only until five did we work]. (12) *Their promise that [only until five will they work] will soon be posted. As is clear, these sentences involve instances of non-reported speech non-finite clauses (10), adjunct clauses (11) and complements of Noun (12). From a formal perspective, Emonds argues that RIDEs project a Discourse Shell (2004, 85). Hence, root and rootlike clauses are analyzed as in (13): (13) XP = Discourse Shell Spec,XP X’ landing site X IP = Discourse Projection of YP DP I’ I VP The phrase which Emonds terms YP corresponds to any constituent undergoing movement to the specifier of the Discourse Shell. Only RIDEs project this Discourse Shell, accounting for the occurrence of root transformations in root-like contexts. However, this analysis poses at least two problems. First, within English any complement 15 NEGATIVE PREPOSING IN COMPLEMENT CLAUSES ATLANTIS. Journal of the Spanish Association of Anglo-American Studies. 40.1 (June 2018): 11-37 • issn 0210-6124 | e-issn 1989-6840 of V or A is assumed to be a RIDE as long as it is finite, thereby allowing an RT. This is contrary to fact, as demonstrated by the data in (14), involving the A surprised: (14) *He was surprised that [never in my life had I seen a hippopotamus]. (Hooper and Thompson 1973, 479) Emonds’s analysis does not apply in Spanish, which allows NPr in non-RIDE environments, alongside other RTs—see Jiménez-Fernández and Miyagawa (2014), Camacho-Taboada and Jiménez-Fernández (2014). One such non-RIDE is the complement of a noun. In this specific type of embedding Spanish produces acceptable sentences as shown in the contrast between (15a) and (15b): (15a) *Their promise that [only until five will they work] will be hard to keep. (Emonds 2004, 77) In (15b) the complement of the noun promesa [“promise”] involves NPr and the result is grammatical, contrary to what Emonds’s approach predicts.3 (15b) La promesa del Gobierno de que [solo hasta las cinco puedan trabajar the promise of the government of that only until the five can-pres.3pl to work los empleados] no ha gustado en Bruselas. the employees not have-pres.3sg liked in Brussels [“The Government’s promise that employees could work only until five hasn’t been very welcome in Brussels.”] Emonds is aware that there are languages that allow RTs to occur in more environments than in English, French and German. To account for these languages, he proposes the Discourse Projection Parameter, which states that “[p]articular languages specify progressively larger classes of finite clauses as Discourse Projections” (Emonds 2004, 82). In other words, the class of RIDEs is larger in certain languages. Emonds’s proposal tells us that there are variations among languages in terms of the root operations that can occur, but it fails to tell us why this variation exists. In this paper I analyze data from English and Spanish concerning the RT negative preposing and account for this parametric variation by proposing that in some languages, such 3 I am assuming, with Emonds (2004) and Haegeman (2000; 2012), that NPr covers fronting of both negative elements such as never alongside non-assertive constituents such as hardly, scarcely, rarely, etc. These non-assertive elements share the same interpretive property of fronted negative elements with regard to the modification of the sentence polarity. The syntactic trait that concerns us here is that both types of elements involve fronting and inversion, which are crucial in my analysis. As for Spanish, Ignacio Bosque (1980, 27) also holds that forms such as quantifiers like poco [“few”], and adverbs like sólo [“only”], raramente [“rarely”], apenas [“scarcely”], etc. are negative or non-assertive. In this work I will combine the grouping of negative and nonassertive forms in the same class. 16 ÁNGEL LUIS JIMÉNEZ-FERNÁNDEZ ATLANTIS. Journal of the Spanish Association of Anglo-American Studies. 40.1 (June 2018): 11-37 • issn 0210-6124 | e-issn 1989-6840 as Spanish, the discourse features that trigger root-type movement operations such as negative preposing are inherited by T, whereas in languages such as English these discourse features remain at C. Feature inheritance has been independently motivated by Chomsky (2008), Richards (2007), Miyagawa (2005; 2010 and 2017) and Jiménez- Fernández (2010; 2011). Here, I implement the theory of feature inheritance by showing that it is ultimately the reason underlying the parametric variation as far as the root phenomenon is concerned. Several works have criticized Emonds’s original treatment of the root phenomenon. In an early work, Hooper and Thompson (1973) identify certain adjunct clauses where NPr is possible: (16) Robert was quite nervous, because never before had he had to borrow money. To account for the distribution of root transformations, Hooper and Thompson argue that emphasis is common to all the root transformations that Emonds lists. Emphasis, in this sense of the term, is most naturally expressed in asserted (and hence non-factive) clauses, where Emonds’s root transformations apply, regardless of whether or not it is a main or subordinate clause. However, as I will discuss below, in languages such as Spanish, presupposed clauses are completely compatible with root transformations. Extending the observation by Hooper and Thompson, Haegeman (2006b; 2010; 2012, 263) and Haegeman and Ürögdi (2010) all argue that there is some sort of movement in non-asserted clauses, and that this movement stops the root-type transformations from applying. As such, the root/non-root distinction is a function not of some fundamental difference in clause-type, but rather is a function of whether movement is allowed to occur in a given context. If a specific type of movement has already taken place, this triggers an intervention effect and blocks other movements, such as NPr, from occurring. In line with Haegeman (2006b, 2010; 2012, 199) and Haegeman and Ürögdi (2010), I adopt this idea that there is an intervention effect that blocks the application of root transformations such as topicalization and negative preposing in non-root environments. The central observation in this paper is that in certain languages, nonroot environments that ought to trigger an intervention effect in fact do not, making it appear that the root context in these languages is distributed more widely, as already observed by Emonds (2004). Jiménez-Fernández and Miyagawa have discussed two information-structure operations that avoid this intervention effect (2014). In contrast to English, these authors claim that clitic left dislocation in Spanish and some types of scrambling in Japanese are allowed in non-root contexts. Extending on this, in this paper I argue that in Spanish the movement in NPr instances is not to Spec-CP, but rather Spec-TP, which is made possible by the relevant feature being inherited by T. NPr does not compete with other types of movement since the landing site will be different. In English, on 17 NEGATIVE PREPOSING IN COMPLEMENT CLAUSES ATLANTIS. Journal of the Spanish Association of Anglo-American Studies. 40.1 (June 2018): 11-37 • issn 0210-6124 | e-issn 1989-6840 the other hand, the discourse feature remains in C and NPr targets Spec-CP. In factive complement clauses NPr is blocked because it competes with the factive operator for the same syntactic position. This suggests that the cross-linguistic variation that Emonds (2004) observes is not explained by the fact that the root context is extended to more environments in some languages, but that the so-called root transformations in these languages apply inside the TP projection, which avoids the intervention effect that would otherwise be triggered by an already existing movement to Spec-CP. The article is organized as follows. In section two, I discuss the syntax and interpretation of NPr in English and Spanish, suggesting that, behind this fronting operation, there is a combination of the features [+focus, +negation] for English and [+emphasis, +negation] for Spanish. Section three addresses the connection between the assertion of complement clauses and RTs, and presents data in support to my view that NPr is compatible with non-asserted clauses in Spanish, while not in English. In section four I argue that the intervention-based analyses proposed by Haegeman and Ürögdi (2010) and Jiménez-Fernández and Miyagawa (2014) can fully account for Spanish RTs, in contrast with English RTs, if feature inheritance from C to T is assumed in embedded contexts. Section five focuses on the methodology that I have used in the experiment with native speakers of the two languages under examination, along with the data that the informants were confronted with, and discuss the results of the survey, highlighting the systematic comparison of negative preposing in English and Spanish. In section six I address the analysis of NPr in the two languages, proposing that discourse features (more precisely, a focus feature) are inherited by T in Spanish to account for NPr, whereas in English they are retained in C. Finally, some concluding remarks are presented in section seven. 2. Describing Negative Preposing in English and Spanish Negative preposing has been defined as a subtype of focus fronting in literature on English—Haegeman (2012, 44), Emonds (2004), De Clercq (2010)—which involves movement of a negative or non-assertive constituent to the left periphery of the sentence, by means of which the polarity of the sentence is affected, yielding a negative sentence. This is illustrated in (17a) and (17b): (17a) [On no account] could she move to Paris. (De Clercq 2010, 231) (17b) [Not a bite] did he eat. (Green 1976, 384) As stated earlier, both adjunct prepositional phrases (PP) and argument determiner phrases (DP) can be fronted in this type of construction. Among the defining properties of English NPr are the subject-auxiliary inversion attested in (17) and an emphasis on the negative polarity of the sentence. It has been claimed that the first trait is common 18 ÁNGEL LUIS JIMÉNEZ-FERNÁNDEZ ATLANTIS. Journal of the Spanish Association of Anglo-American Studies. 40.1 (June 2018): 11-37 • issn 0210-6124 | e-issn 1989-6840 to all types of focus fronting, which has led linguists to argue in favor of an analysis whereby NPr is triggered by a focus feature. However, there are clear cases of focus fronting in English which do not include subject-auxiliary inversion: (18) THIS BOOK I don’t need (but that one I do). (Haegeman 2012, 8) Note that in NPr if no subject-auxiliary inversion takes place, the result is completely ungrammatical: (19a) *[On no account] she could move to Paris. (19b) *[Not a bite] he ate. The impossibility of the non-inverted version shows that the sentences in (17) are not instances of topicalization, which requires the order subject+auxiliary illustrated in (20): (20) This book you should read. (Hooper and Thompson 1973, 468) The second property that describes the phenomenon under study is an emphasis on the negative polarity of the sentence in question caused by the fronting of the negative constituent. In this respect, sentence negation is involved in NPr. Haegeman (2000) and De Clercq (2010) argue that if a negative tag can be added to a sentence it is because this particular sentence is negative. If this is correct and sentences with NPr are negative, cases of NPr are expected to accept a neither-tag. This prediction is borne out in light of the data in (21): (21) On no account could she move to Paris, and neither could Jane. (adapted from De Clercq 2010, 232) In addition, sentence negation is confirmed to be involved in NPr since other types of tags are also allowed: (22) On no account could she move to Paris, could she? Haegeman (2012, 33) provides an analysis of NPr based on the movement of the negative constituent to a designated focus phrase in the left periphery. This movement is triggered by a focus feature. Since what is emphasized is the negative polarity, I assume for English that this focus feature also includes a negative feature. In other words, the triggering feature will be [+focus, +negation], following the spirit of decomposition of discourse categories in terms of features proposed in Jiménez-Fernández (2015). 19 NEGATIVE PREPOSING IN COMPLEMENT CLAUSES ATLANTIS. Journal of the Spanish Association of Anglo-American Studies. 40.1 (June 2018): 11-37 • issn 0210-6124 | e-issn 1989-6840 Next I turn to Spanish. Negative preposing in Spanish has not been studied in depth, although Bosque mentions the existence of a type of fronting which induces negative polarity (1980, 34). He calls these fronting operations “anteposiciones negativas” [“negative preposing”], and they are illustrated in (23a) to (23d): (23a) En modo alguno se puede tolerar tal actitud. in way some SE can-pres.3sg to.tolerate such attitude [“By no means can such an attitude be tolerated.”] (23b) En la/mi vida he estado aquí. in the/my life have-pres.1sg been here [“Never in my life have I been here.”] (23c) De nada carece don Agapito. of nothing lack-pres.3sg don Agapito [“Nothing does Don Agapito lack.”] (23d) De ninguno de esos problemas trató la reunión. of none of those problems treat-past.3sg the meeting [“None of those problems did the meeting discuss.”] (Bosque 1980, 34-35) These are cases of adjunct NPr, which clearly induce a negative interpretation of the sentence polarity, given the corresponding sentences in (24a) to (24d) with no fronting but with the explicit occurrence of the negative adverb no [“not”]:4 (24a) No se puede tolerar tal actitud en modo alguno. not SE can-pres.3sg to.tolerate such attitude in way some [“Such an attitude cannot be tolerated by any means.”]5 4 One of the properties which are highlighted for NPr in both English and Spanish is that the original sentences with no fronting and no negative adverb are ungrammatical (De Clercq 2010; Bosque 1980, 35). Sentences in (i) and (ii) are the corresponding non-fronting sentences in English and Spanish respectively: (i) *She could move to Paris on no account. (ii) *Se puede tolerar tal actitud en modo alguno. SE can-pres.3sg to.tolerate such attitude in way some [“*Such an attitude can be tolerated by no means.”] This can be taken as evidence that fronting makes the sentence polarity negative. In the absence of fronting, some other strategy such as insertion of the negative adverb must be applied. This rule is known as Neg-shift. For the different views in the discussion of this rule see Bosque (1980), Haegeman (2000), Zeijlstra (2004), Tubau (2008), De Clercq (2010). 5 Note that the English translation includes the negative adverb not, which activates the use of nonassertive forms such as any in English—see Klima (1964) for the licensing of these non-assertive forms and their connection with negative polarity. Conversely, in Spanish once the negative adverb no [“not”] occurs in the sentence, the so-called negative concord ensures that all forms agree in their negative value—see Tubau (2008) for discussion and references. 26 ÁNGEL LUIS JIMÉNEZ-FERNÁNDEZ ATLANTIS. Journal of the Spanish Association of Anglo-American Studies. 40.1 (June 2018): 11-37 • issn 0210-6124 | e-issn 1989-6840 (35a) *John regretted that never had he seen Gone with the Wind. (Authier 1992, 334 [example 10b]) (35b) It is true that never in his life has he had to borrow money. (Hooper and Thompson 1973, 476 [example 68]) Haegeman’s analysis (2012) predicts this contrast in that the predicate true is nonreferential (and hence asserted) and as such selects a CP which is not endowed with an event operator. Since it has no operator there is no intervention effect with other possible movements such as topic fronting and NPr in English. On the other hand, regret is factive/referential and hence its sentential complement carries an event operator whose movement blocks any other movement. As illustrated earlier, in languages such as Spanish this intervention analysis does not hold since it is expected that in factive/non-asserted contexts NPr is also blocked, contrary to fact. I provide an explanation based on an implementation of Haegeman’s intervention effects and feature inheritance, a strategy I will now discuss before presenting my proposal. 4.2. Feature inheritance and RTs Based on Chomsky’s notion of feature inheritance (2008), Miyagawa (2005; 2010) has proposed that topics in the form of scrambling move to Spec-TP in languages such as Japanese because discourse features, which start out in C, may be inherited by T. T triggers movement of topics to Spec-TP. This is the reason why scrambling is compatible with all types of predicate classes in Hooper and Thompson’s study. This is exemplified for Japanese in (36): (36) John-wa [konohon-o zibun-no-kodomo-gayonda koto]-o kookaisita. John-top this book-acc self’s child-nom read CFACT -acc regret [“John regrets that this book, his child read.”] [Class D] (adapted from Maki, Kaiser and Ochi 1999, 9) The verb regret in Japanese is thoroughly compatible with topic fronting in the form of scrambling, although, as non-asserted, its complement contains an event operator. Jiménez-Fernández and Miyagawa explain this variation in terms of feature inheritance (2014, 283). If the discourse feature triggering movement of the DP kono hono [“this book”] is lowered onto T, the scrambled constituent will move to Spec-TP, not to Spec- CP. Thus, the intervention or competition with the event operator is avoided and the construction is predicted to be acceptable. Jiménez-Fernández and Miyagawa (2014)—after Miyagawa (2010, 89) and Jiménez-Fernández (2010)—make a typological classification of languages depending 27 NEGATIVE PREPOSING IN COMPLEMENT CLAUSES ATLANTIS. Journal of the Spanish Association of Anglo-American Studies. 40.1 (June 2018): 11-37 • issn 0210-6124 | e-issn 1989-6840 on the kind of grammatical features inherited by T. Languages can be grouped into three types: agreement-based languages, discourse-configurational languages—see also Kiss (1995)—and languages that are both agreement-oriented and discourseprominent. This typology—adapted from Jiménez-Fernández (2010, 32; 2011, 15; see also Miyagawa [2010, 89])—is outlined below: (a) Cφ, δ → Tδ … discourse-prominent, e.g., Japanese, Korean, etc. [φ: agreement features, δ: discourse features] (b) Cφ, δ → Tφ … agreement-prominent, e.g., English and most Indo-European languages. (c) Cφ, δ → Tφ, δ …discourse-prominent, agreement-oriented, e.g., Spanish, Turkish, Greek, etc. From this feature-based typology we can infer that a discourse feature such as the topic feature is inherited by T; the EPP under T triggers movement of the probed category to Spec-TP (Jiménez-Fernández 2010, 33-34). From this perspective, discourse-configurational languages are expected not to show any intervention effect for RTs. This is the case for scrambled topics in Japanese, as shown above. The question that the intervention-based approach poses is whether the same phenomenon is attested in some form in all languages. English seems to retain discoursefeatures in C, which supports Haegeman’s view that in this language non-asserted clauses display intervention effects (2010) , since both discourse-type categories and an operator compete for Spec-CP. This is illustrated in (37): (37) [CP OPi Cevent+δ . . . [FP ti [TP T [vP DP v+V DP]]] Movement of any material to Spec-CP from vP is blocked by the intervener OP. In the absence of operator movement (as in asserted clauses), there is no intervention and hence topics can be fronted to Spec-CP: (38) [CP TOPi Cδ [TP T [vP DP v+V DPi]]] A rather different case in point is Spanish. In clear contrast to English, Spanish is discourse-configurational—additionally, Spanish is also agreement-based. This is interpreted by Jiménez-Fernández and Miyagawa in terms of lowering δ-features onto T (2014, 292). If this be the case, it has an important consequence for the intervention effect shown by operator movement in non-asserted CPs. If CLLD-ed [φ: agreement features, δ: discourse features] (b) Cφ, δ → Tφ … agreement-prominent, e.g., English and most Indo-European languages. (c) Cφ, δ → Tφ, δ …discourse-prominent, agreement-oriented, e.g., Spanish, Turkish, Greek, etc. From this feature-based typology we can infer that a discourse feature such as the topic feature is inherited by T; the EPP under T triggers movement of the probed category to Spec- TP (Jiménez-Fernández 2010, 33-34). From this perspective, discourse-configurational languages are expected not to show any intervention effect for RTs. This is the case for scrambled topics in Japanese, as shown above. The question that the intervention-based approach poses is whether the same phenomenon is attested in some form in all languages. English seems to retain discourse-features in C, which supports Haegeman’s view that in this language non-asserted clauses display intervention effects (2010) , since both discourse-type categories and an operator compete for Spec-CP. This is illustrated in (37): (37) [CP OPi Cevent+δ . . . [FP ti [TP T [vP DP v+V DP]]] Movement of any material to Spec-CP from vP is blocked by the intervener OP. In the absence of operator movement (as in asserted clauses), there is no intervention and hence topics can be fronted to Spec-CP: (38) [CP TOPi Cδ [TP T [vP DP v+V DPi]]] A rather different case in point is Spanish. In clear contrast to English, Spanish is discourse-configurational—additionally, Spanish is also agreement-based. This is interpreted by Jiménez-Fernández and Miyagawa in terms of lowering δ-features onto T (2014, 292). If this be the case, it has an important consequence for the intervention effect shown by operator movement in non-asserted CPs. If CLLD-ed topics and scrambled constituents are moved to Spec-TP, operator movement does not interfere with any subsequent movement in the same construction. This predicts that irrespective of the asserted or non-asserted character of CPs, topics moving to Spec-TP are completely compatible with operator movement in referential [φ: agreement features, δ: discourse features] (b) Cφ, δ → Tφ … agreement-prominent, e.g., English and most Indo-European languages. (c) Cφ, δ → Tφ, δ …discourse-prominent, agreement-oriented, e.g., Spanish, Turkish, Greek, etc. From this feature-based typology we can infer that a discourse feature such as the topic feature is inherited by T; the EPP under T triggers movement of the probed category to Spec- TP (Jiménez-Fernández 2010, 33-34). From this perspective, discourse-configurational languages are expected not to show any intervention effect for RTs. This is the case for scrambled topics in Japanese, as shown above. The question that the intervention-based approach poses is whether the same phenomenon is attested in some form in all languages. English seems to retain discourse-features in C, which supports Haegeman’s view that in this language non-asserted clauses display intervention effects (2010) , since both discourse-type categories and an operator compete for Spec-CP. This is illustrated in (37): (37) [CP OPi Cevent+δ . . . [FP ti [TP T [vP DP v+V DP]]] Movement of any material to Spec-CP from vP is blocked by the intervener OP. In the absence of operator movement (as in asserted clauses), there is no intervention and hence topics can be fronted to Spec-CP: (38) [CP TOPi Cδ [TP T [vP DP v+V DPi]]] A rather different case in point is Spanish. In clear contrast to English, Spanish is discourse-configurational—additionally, Spanish is also agreement-based. This is interpreted by Jiménez-Fernández and Miyagawa in terms of lowering δ-features onto T (2014, 292). If this be the case, it has an important consequence for the intervention effect shown by operator movement in non-asserted CPs. If CLLD-ed topics and scrambled constituents are moved to Spec-TP, operator movement does not interfere with any subsequent movement in the same construction. This predicts that irrespective of the asserted or non-asserted character of CPs, topics moving to Spec-TP are completely compatible with operator movement in referential 28 ÁNGEL LUIS JIMÉNEZ-FERNÁNDEZ ATLANTIS. Journal of the Spanish Association of Anglo-American Studies. 40.1 (June 2018): 11-37 • issn 0210-6124 | e-issn 1989-6840 topics and scrambled constituents are moved to Spec-TP, operator movement does not interfere with any subsequent movement in the same construction. This predicts that irrespective of the asserted or non-asserted character of CPs, topics moving to Spec-TP are completely compatible with operator movement in referential CPs. The prediction is borne out by scrambling in Japanese—see (39)—and contrastive and familiar topics in Spanish—see (40).9 (39) [CP OPi Cevent+δ [FP ti [TP TOPj Tδ [vP DP v+V DPj]]] (40) Siento que el artículo no lo hayan publicado regret-pres.1sg that the article not CL have-pres.subj.3sg published en esa revista. in that journal [“I regret that the article hasn’t been published in that journal.”] The derivation shown in (39) is based on the inheritance of δ-features from C to T. The two arrows never coincide, avoiding any blocking effect, and predicting that a sentence such as (40) is completely well-formed. Now the question arises as to whether the same analysis can be proposed for negative preposing, a phenomenon that does not raise any doubts as to whether there is movement or not. Therefore, in what follows I explore the connection between NPr and assertedness/factivity, based on data from an experiment carried out among native speakers of English and Spanish. 9 Two points are in need of clarification here. The first concerns the discourse-prominent character of Spanish. According to Kiss, a language is discourse-configurational if it arranges the sentence elements depending on the discourse (information structure) function of those elements (1995, 5). Spanish clearly complies with this premise in that the organization of the sentence hinges on the discourse role of its members. On the opposite side, we find English, which arranges its sentences according to the syntactic function and position of the elements involved. In the minimalist system that I adopt here, being discourse-prominent means that at least some discourse features are inherited by T from C. And this is the second point to be clarified. As far as CLLD, I have argued elsewhere that familiar topics and contrastive topics target Spec-TP after lowering of topic features to T (Jiménez- Fernández 2010). Evidence in favor of this analysis comes from binding. Anaphors are bound from an argument position (typically Spec-TP), which does not allow reconstruction. This is precisely what we find in CLLD: (vi) A Ángela la buscó su propio padre por todos lados. to Angela CL search-past.3sg her own father by all sides [“Angela was looked for everywhere by her own father.”] If the DP a Ángela had moved to a position in CP, it would allow reconstruction and then the bound reading could not be obtained. On the contrary, if this DP moves to Spec-TP, it can perfectly bind the anaphor in the DP subject su propio padre. Accordingly, topics may move to Spec-TP in Spanish, and this is allowed by discoursefeature inheritance. CPs. The prediction is borne out by scrambling in Japanese—see (39)—and contrastive and familiar topics in Spanish—see (40).9 (39) [CP OPi Cevent+δ [FP ti [TP TOPj Tδ [vP DP v+V DPj]]] (40) Siento que el artículo no lo hayan publicado regret-PRES.1SG that the article not CL have-PRES.SUBJ.3SG published en esa revista. in that journal [“I regret that the article hasn’t been published in that journal.”] The derivation shown in (39) is based on the inheritance of δ-features from C to T. The two arrows never coincide, avoiding any blocking effect, and predicting that a sentence such as (40) is completely well-formed. Now the question arises as to whether the same analysis can be proposed for negative preposing, a phenomenon that does not raise any doubts as to whether there is movement or not. Therefore, in what follows I explore the connection between NPr and assertedness/factivity, based on data from an experiment carried out among native speakers of English and Spanish. 9 Two points are in need of clarification here. The first concerns the discourse-prominent character of Spanish. According to Kiss, a language is discourse-configurational if it arranges the sentence elements depending on the discourse (information structure) function of those elements (1995, 5). Spanish clearly complies with this premise in that the organization of the sentence hinges on the discourse role of its members. On the opposite side, we find English, which arranges its sentences according to the syntactic function and position of the elements involved. In the minimalist system that I adopt here, being discourse-prominent means that at least some discourse features are inherited by T from C. And this is the second point to be clarified. As far as CLLD, I have argued elsewhere that familiar topics and contrastive topics target Spec-TP after lowering of topic features to T (Jiménez-Fernández 2010). Evidence in favor of this analysis comes from binding. Anaphors are bound from an argument position (typically Spec-TP), which does not allow reconstruction. This is precisely what we find in CLLD: (vi) A Ángela la buscó su propio padre por todos lados. to Angela CL search-PAST.3SG her own father by all sides [“Angela was looked for everywhere by her own father.”] If the DP a Ángela had moved to a position in CP, it would allow reconstruction and then the bound reading could not be obtained. On the contrary, if this DP moves to Spec-TP, it can perfectly bind the anaphor in the DP subject su propio padre. Accordingly, topics may move to Spec-TP in Spanish, and this is allowed by discourse-feature inheritance. 29 NEGATIVE PREPOSING IN COMPLEMENT CLAUSES ATLANTIS. Journal of the Spanish Association of Anglo-American Studies. 40.1 (June 2018): 11-37 • issn 0210-6124 | e-issn 1989-6840 5. The Experiment, Data and Statistics As stated at the beginning of the present work, my working hypothesis is that in English an RT such as NPr is compatible with asserted verb classes but not with nonasserted verb classes, whereas in Spanish NPr may occur in sentential complements of all types of verbs. 5.1. Methodology In order to check the working hypothesis and verify the predictions discussed in the previous sections, a test has been built for a systematic interpretation analysis concerning the acceptability of NPr in the sentential complements of the different verb classes identified by Hooper and Thompson (1973) in the two languages under examination, English and Spanish. The main purpose of the experiment is to obtain data from a real use of the language, as opposed to the judgments found in purely theoretical approaches. In fact, NPr of an adjunct is judged as grammatical in the complement of all verbal classes by Haegeman (2012, 39), whereas Hooper and Thompson (1973, 479) and Green (1976, 388) deem the same construction unacceptable. In addition, in Spanish no study exists on the possible asymmetry between argument and adjunct fronting. The results of this survey are intended to fill these gaps. In the tests, informants were asked to express their judgments as either OK, to indicate “full acceptance or grammaticality,”?? for “marginal, but acceptable” constructions and NO in case of “unacceptable or ungrammatical” ones. The sentences were conveniently randomized so that the informants could not create patterns of behavior. Eighty-nine data sets for Spanish and seventy-six for English have been collected, taking into consideration only full responses. The Spanish group was made up of fourth-year students from the English Studies degree at the University of Seville, while the English subjects were students of a degree in Linguistics at the University of Cambridge (UK). 5.2. Data for NPr in complement clauses The survey contained two sentences for each verbal class in Hooper and Thompson’s classification (1973), both for English and for Spanish. The sentences contained either argument fronting or adjunct fronting in the form of NPr. In the survey, I provided two different sentences per verbal class for each type of NPr, namely argument and adjunct NPr. Below is a sample of the sentences tested; capitals letters are used to indicate that informants should take the preposed element as emphatic, although emphasis does not involve contrast: 1. NPr in English: Argument fronting (based on Radford 2009, 327) (41a) He said that NO OTHER COLLEAGUE would he turn to. [Class A] (41b) I guess that NO OTHER COLLEAGUE could he turn to. [Class B] (41c) I doubt that NO OTHER COLLEAGUE could he turn to. [Class C] 30 ÁNGEL LUIS JIMÉNEZ-FERNÁNDEZ ATLANTIS. Journal of the Spanish Association of Anglo-American Studies. 40.1 (June 2018): 11-37 • issn 0210-6124 | e-issn 1989-6840 (41d) I was surprised that NO OTHER COLLEAGUE could he turn to. [Class D] (41e) I realized that NO OTHER COLLEAGUE could he turn to. [Class E] 2. NPr in English: Adjunct fronting (based on Meinunger 2004, 215) (42a) He said that NEVER IN HIS LIFE did he do anything like that. [Class A] (42b) It appears that NEVER IN HIS LIFE did he do anything like that. [Class B]: (42c) John denied that NEVER IN HIS LIFE had he seen this woman. [Class C] (42d) I resent that NEVER IN MY LIFE did I do anything like that. [Class D] (42e) I found out that NEVER BEFORE had he had to borrow money. [Class E] (Hooper and Thompson 1973, 480) 3. Spanish NPr: Argument fronting (43a) Dijo que A NINGÚN OTRO COMPAÑERO podría acudir Juan en busca de ayuda. [Class A] (43b) Supongo que A NINGÚN OTRO COMPAÑERO podía acudir Juan en busca de ayuda. [Class B] (43c) Dudo de que A NINGÚN OTRO COMPAÑERO pudiera acudir Juan en busca de ayuda. [Class C] (43d) Me sorprende que A NINGÚN OTRO COMPAÑERO pueda acudir Juan en busca de ayuda. [Class D] (43e) En seguida noté que A NINGÚN OTRO COMPAÑERO podría acudir Juan en busca de ayuda. [Class E] 4. NPr in Spanish: Adjunct fronting (44a) Dijo que NUNCA EN SU VIDA podría hacer Juan algo. [Class A] (44b) Parece que NUNCA EN SU VIDA había hecho Juan algo parecido. [Class B] (44c) Dudo de que NUNCA EN SU VIDA haya visto Juan a esa mujer. [Class C] (44d) Me sorprende que NUNCA EN SU VIDA haya hecho Juan algo parecido. [Class D] (44e) Descubrí que NUNCA EN SU VIDA había hecho Juan algo así. [Class E] 5.3. Discussion of results In this section, I am reporting on the results of the experiment with the purpose of finding empirical evidence for the hypothetical difference between English and Spanish and for the distinction between arguments and adjuncts with respect to NPr. In table 2 the results from the English test are presented.10 10 To avoid unnecessary complications with decimals, in both tables I have rounded all figures. Also it must be clear that only fully acceptable responses were considered, so marginal and ungrammatical sentences are excluded. 31 NEGATIVE PREPOSING IN COMPLEMENT CLAUSES ATLANTIS. Journal of the Spanish Association of Anglo-American Studies. 40.1 (June 2018): 11-37 • issn 0210-6124 | e-issn 1989-6840 Table 2. English Negative Preposing: OK Responses English NPr Class A Class B Class C Class D Class E Argument NPr 85%60%0%0%65% Adjunct NPr 90%80%15%25%75% In general, informants judged NP in asserted contexts—classes A, B and E—as grammatical, albeit a lower figure was obtained for class B in argument NPr and class E in adjunct NPr. However, the results are not low enough to consider the relevant sentences ill-formed. Note, on the other hand, that classes C and D yield degraded results. No speaker has rated argument NPr as well-formed, and extremely low figures (15% and 25%) were achieved for adjunct NPr in classes C and D. This shows that arguments and adjuncts exhibit similar behavior with respect to NPr, hence Haegeman’s claim that there is asymmetry between arguments and adjuncts with respect to fronting is not supported by NPr (2012, 39). In both argument and adjunct NPr there is movement to the left periphery, which is confirmed by the intervention effect detected in classes C and D for both types of NPr tested. In the case of Spanish, the results are shown in table 3: Table 3. Spanish Negative Preposing: OK responses Spanish NPr Class A Class B Class C Class D Class E Argument NPr 70%80%85%70%75% Adjunct NPr 75%90%75%85%80% Table 3 clearly shows that NPr in Spanish is available in all classes regardless of whether the fronting involves an argument or an adjunct. When observed as a whole, there are some differences in terms of acceptability between argument and adjunct NPr in that scores for the former are a bit lower than those obtained for adjunct NPr. However, the difference is not sufficient to consider them as significantly distinct. This confirms my hypothesis that in Spanish there is no intervention, precisely because NPr and operator movement do not interfere with each other in the derivation, which supports an analysis where NPr in Spanish is movement to Spec-TP. On the other hand, the English results show that NPr is only acceptable in asserted clauses, i.e., those selected by verbs of classes A, B and E. This supports the view that in these sentential complements operator movement intervenes with other movements, thereby blocking NPr. 32 ÁNGEL LUIS JIMÉNEZ-FERNÁNDEZ ATLANTIS. Journal of the Spanish Association of Anglo-American Studies. 40.1 (June 2018): 11-37 • issn 0210-6124 | e-issn 1989-6840 6. The Analysis of NPr and the Opposition Spec-CP vs. Spec-TP The conclusion from my previous section is that Spanish is much more flexible than English with respect to the availability of NPr in asserted/non-asserted clauses. In addition, the data confirm my claim that different syntactic strategies can be used by different languages to express a similar meaning. This leads me to propose that the feature inheritance system presented in subsection 3.1 could be the basis for the variation in NPr between English and Spanish, similar to topic fronting in the two languages. By extending the analysis presented earlier, involving feature inheritance and intervention, the data from NPr can be accommodated perfectly. For English, I suggest that, since the discourse feature responsible for triggering movement of the negative constituent in NPr is retained in C, both arguments and adjuncts undergo movement to Spec-CP. If NPr takes place in a root clause or an asserted embedded clause—a non- RIDE in Emonds’s terms (2004, 77)—the result is well-formed since there is no event operator intervening. Either a DP argument, a PP or AdvP adjunct will target Spec-CP, valuing the features [+focus, +negation]: (45) [CP DPi/PPi Cδ [TP T [vP DP v+V DPi/PPi]]] In non-asserted contexts (RIDEs), an event operator moves to CP and this blocks any other movement, leading to the prediction that NPr will be banned since the features under C remain unvalued: (46) [CP OPi Cevent+δ . . . [FP ti [TP T [vP DP v+V DP/PP]]]] Conversely, in Spanish the discourse feature is lowered onto T, and hence any discourse movement targets Spec-TP. This predicts that in non-asserted clauses, where there is operator movement, NPr and other movements should not be mutually exclusive, as illustrated and discussed above in subsection 4.3. The derivation that I suggest for Spanish NPr in non-asserted contexts is as follows: (47) [CP OPi Cevent+δ [FP ti [TP DPj/PPj Tδ [vP DP v+V DPj/PPj]]]] As shown by the two arrows, there is no interference between the two movements, so no intervention is displayed and the result is a grammatical sentence. Note that the discourse features have been inherited by T. In this configuration the negative differences in terms of acceptability between argument and adjunct NPr in that scores for the former are a bit lower than those obtained for adjunct NPr. However, the difference is not sufficient to consider them as significantly distinct. This confirms my hypothesis that in Spanish there is no intervention, precisely because NPr and operator movement do not interfere with each other in the derivation, which supports an analysis where NPr in Spanish is movement to Spec-TP. On the other hand, the English results show that NPr is only acceptable in asserted clauses, i.e., those selected by verbs of classes A, B and E. This supports the view that in these sentential complements operator movement intervenes with other movements, thereby blocking NPr. 6. THE ANALYSIS OF NPR AND THE OPPOSITION SPEC-CP VS. SPEC-TP The conclusion from my previous section is that Spanish is much more flexible than English with respect to the availability of NPr in asserted/non-asserted clauses. In addition, the data confirm my claim that different syntactic strategies can be used by different languages to express a similar meaning. This leads me to propose that the feature inheritance system presented in subsection 3.1 could be the basis for the variation in NPr between English and Spanish, similar to topic fronting in the two languages. By extending the analysis presented earlier, involving feature inheritance and intervention, the data from NPr can be accommodated perfectly. For English, I suggest that, since the discourse feature responsible for triggering movement of the negative constituent in NPr is retained in C, both arguments and adjuncts undergo movement to Spec-CP. If NPr takes place in a root clause or an asserted embedded clause—a non-RIDE in Emonds’s terms (2004, 77)—the result is well-formed since there is no event operator intervening. Either a DP argument, a PP or AdvP adjunct will target Spec-CP, valuing the features [+focus, +negation]: (45) [CP DPi/PPi Cδ [TP T [vP DP v+V DPi/PPi]]] In non-asserted contexts (RIDEs), an event operator moves to CP and this blocks any other movement, leading to the prediction that NPr will be banned since the features under C remain unvalued: (46) [CP OPi Cevent+δ . . . [FP ti [TP T [vP DP v+V DP/PP]]]] differences in terms of acceptability between argument and adjunct NPr in that scores for the former are a bit lower than those obtained for adjunct NPr. However, the difference is not sufficient to consider them as significantly distinct. This confirms my hypothesis that in Spanish there is no intervention, precisely because NPr and operator movement do not interfere with each other in the derivation, which supports an analysis where NPr in Spanish is movement to Spec-TP. On the other hand, the English results show that NPr is only acceptable in asserted clauses, i.e., those selected by verbs of classes A, B and E. This supports the view that in these sentential complements operator movement intervenes with other movements, thereby blocking NPr. 6. THE ANALYSIS OF NPR AND THE OPPOSITION SPEC-CP VS. SPEC-TP The conclusion from my previous section is that Spanish is much more flexible than English with respect to the availability of NPr in asserted/non-asserted clauses. In addition, the data confirm my claim that different syntactic strategies can be used by different languages to express a similar meaning. This leads me to propose that the feature inheritance system presented in subsection 3.1 could be the basis for the variation in NPr between English and Spanish, similar to topic fronting in the two languages. By extending the analysis presented earlier, involving feature inheritance and intervention, the data from NPr can be accommodated perfectly. For English, I suggest that, since the discourse feature responsible for triggering movement of the negative constituent in NPr is retained in C, both arguments and adjuncts undergo movement to Spec-CP. If NPr takes place in a root clause or an asserted embedded clause—a non-RIDE in Emonds’s terms (2004, 77)—the result is well-formed since there is no event operator intervening. Either a DP argument, a PP or AdvP adjunct will target Spec-CP, valuing the features [+focus, +negation]: (45) [CP DPi/PPi Cδ [TP T [vP DP v+V DPi/PPi]]] In non-asserted contexts (RIDEs), an event operator moves to CP and this blocks any other movement, leading to the prediction that NPr will be banned since the features under C remain unvalued: (46) [CP OPi Cevent+δ . . . [FP ti [TP T [vP DP v+V DP/PP]]]] Conversely, in Spanish the discourse feature is lowered onto T, and hence any discourse movement targets Spec-TP. This predicts that in non-asserted clauses, where there is operator movement, NPr and other movements should not be mutually exclusive, as illustrated and discussed above in subsection 4.3. The derivation that I suggest for Spanish NPr in nonasserted contexts is as follows: (47) [CP OPi Cevent+δ [FP ti [TP DPj/PPj Tδ [vP DP v+V DPj/PPj]]]] As shown by the two arrows, there is no interference between the two movements, so no intervention is displayed and the result is a grammatical sentence. Note that the discourse features have been inherited by T. In this configuration the negative constituent values the [+emphasis, +negation] features in T. In conclusion, contrary to English non-RIDEs, in Spanish non-RIDEs NPr is available because its movement does not interfere with movement of the operator. Empirical evidence for positing that Spanish NPr involves movement to Spec-TP comes from principle A of the Binding Theory. It is generally assumed that anaphors are bound (hence c-commanded) by an antecedent in their local domain—the TP containing both the anaphor and its binder (see Chomsky 2008). This means that the antecedent must sit in an argument position (A-position). Spec-CP is a non-argument position (A’-position), whereas Spec-TP is an A-position. Spec-TP may create a new binding configuration (Lasnik 2003, 94; Jiménez-Fernández and Miyagawa 2014; Miyagawa 2010, 66-68; 2017, 77-78). If the fronted constituent in NPr can be an antecedent for an anaphor in a lower position, this means that the emphatic operator has moved to an A-position. The prediction is borne out, as illustrated by the following contrast: (48) [A ninguna de las crías]i les daba su propiai madre to none of the offspring them give-PAST.3SG their own mother de comer. of to.eat [“None of the offspring were fed by their own mother.”] (49) *Su propiai madre no les daba de comer [a ninguna de las crías]i. 33 NEGATIVE PREPOSING IN COMPLEMENT CLAUSES ATLANTIS. Journal of the Spanish Association of Anglo-American Studies. 40.1 (June 2018): 11-37 • issn 0210-6124 | e-issn 1989-6840 constituent values the [+emphasis, +negation] features in T. In conclusion, contrary to English non-RIDEs, in Spanish non-RIDEs NPr is available because its movement does not interfere with movement of the operator. Empirical evidence for positing that Spanish NPr involves movement to Spec-TP comes from principle A of the Binding Theory. It is generally assumed that anaphors are bound (hence c-commanded) by an antecedent in their local domain—the TP containing both the anaphor and its binder (see Chomsky 2008). This means that the antecedent must sit in an argument position (A-position). Spec-CP is a non-argument position (A’-position), whereas Spec-TP is an A-position. Spec-TP may create a new binding configuration (Lasnik 2003, 94; Jiménez-Fernández and Miyagawa 2014; Miyagawa 2010, 66-68; 2017, 77-78). If the fronted constituent in NPr can be an antecedent for an anaphor in a lower position, this means that the emphatic operator has moved to an A-position. The prediction is borne out, as illustrated by the following contrast: (48) [A ninguna de las crías]i les daba su propiai madre to none of the offspring them give-past.3sg their own mother de comer. of to.eat [“None of the offspring were fed by their own mother.”] (49) *Su propiai madre no les daba de comer [a ninguna de las crías]i. In (49) there is no movement of the negative constituent. Instead, it is the subject containing the anaphor that moves to Spec-TP. In this configuration the anaphor is unbound by its antecedent, yielding an ill-formed sentence. Conversely, in (48) the negative antecedent undergoes movement to Spec-TP and binds the anaphor, thereby satisfying principle A of the Binding Theory. Binding can only be obtained if the antecedent occupies Spec-TP, providing evidence that NPr targets Spec-TP in Spanish. 7. Final Remarks This paper demonstrates that root transformations such as negative preposing in Spanish occur much more freely than the equivalents in English. This variation points to the fact that a semantic account for the occurrence of NPr in complement clauses is not accurate. It may also be a challenge to the operator-movement approach to factive/ non-asserted clauses proposed by Haegeman (2010), Haegeman and Ürögdi (2010) and Haegeman (2012, 267) . However, I have argued that, in line with Jiménez-Fernández and Miyagawa (2014), for CLLD and scrambling, NPr involves movement to Spec-CP in English, but to Spec-TP in Spanish. Thus, the operator-movement analysis remains intact. The distribution of NPr (and root transformations in general) correlate with 34 ÁNGEL LUIS JIMÉNEZ-FERNÁNDEZ ATLANTIS. Journal of the Spanish Association of Anglo-American Studies. 40.1 (June 2018): 11-37 • issn 0210-6124 | e-issn 1989-6840 whether discourse features are inherited from C to T. If they are inherited, NPr takes place within TP as in Spanish, and no intervention occurs, but if they stay at C, as in English, NPr competes with an operator movement to Spec-CP, and intervention effects arise. Works Cited Aboh, Enoch. 2005. “Deriving Relative and Factive Constructions in Kwa.” In Contributions to the Thirtieth Incontro di Grammatica Generativa, edited by Laura Brugè, Giuliana Giusti, Nicola Munaro, Walter Schweikert and Giuseppina Turano, 265- 285. Venezia: Libreria Editrice Cafoscarina. Authier, Jean-Marc. 1992. “Iterated CPs and Embedded Topicalisation.” Linguistic Inquiry 23: 329-336. Batllori, Montserrat and Maria Lluïsa Hernanz. 2014. “Variation in the Left Sentence Periphery: Weak Fronted Foci in Spanish and Catalan.” In Syntax over Time: Lexical, Morphological and Information-Structural Interactions, edited by Theresa Biberauer and George Walkden, 280-298. Oxford: Oxford UP. Bianchi, Valentina. 2000. “On Time Adverbials.” Italian Journal of Linguistics 12 (1): 77-106. Bianchi, Valentina and Mara Frascarelli. 2010. “Is Topic a Root Phenomenon?” Iberia: An International Journal of Theoretical Linguistics 2 (1): 43-88. Bosque, Ignacio. 1980. Sobre la negación. Madrid: Cátedra. Camacho-Taboada, Victoria and Ángel L. Jiménez-Hernández. 2014. “Focus Fronting and Root Phenomena in Spanish and English.” In Language Use and Linguistic Structure. Proceedings of the Olomouc Linguistics Symposium, 2013. Volume 3, edited by Joseph Emonds and Markéta Janebová, 47-60. Olomouc: Palacký UP. Chomsky, Noam. 2008. “On Phases.” In Foundational Issues in Linguistic Theory, edited by Robert Freidin, Carlos Otero and Maria-Luisa Zubizarreta, 133-166. Cambridge, MA: The MIT Press. Cinque, Guglielmo. 1990. Types of A’-Dependencies. Cambridge, MA: The MIT Press. De Clercq, Karen. 2010. “NEG-shift in English: Evidence from PP-adjuncts.” In Proceedings of the 12th Seoul International Conference on Generative Grammar: 2010 Movement in Minimalism, 231-251, edited by Duh-Ho An and Joo-Tech Kim. Seoul: Hankuk. de Cuba, Carlos and Barbara Ürögdi. 2009. “Eliminating Factivity from Syntax: Sentential Complements in Hungarian.” In Approaches to Hungarian. Volume 11: Papers from the 2007 New York Conference, edited by Marcel den Dikken and Robert M. Vago, 29-65. Amsterdam and Philadelphia, PA: John Benjamins. Emonds, Joseph. 1969. “Root and Structure-Preserving Transformations.” PhD diss., Massachusetts Institute of Technology. ––. 1976. A Transformational Approach to English Syntax. New York: Academic Press. 35 NEGATIVE PREPOSING IN COMPLEMENT CLAUSES ATLANTIS. Journal of the Spanish Association of Anglo-American Studies. 40.1 (June 2018): 11-37 • issn 0210-6124 | e-issn 1989-6840 ––. 2004. “Unspecified Categories as the Key to Root Constructions.” In Peripheries: Syntactic Edges and their Effects, edited by David Adger, Cecile De Cat and George Tsoulas, 75-121. Dordrecht: Kluwer. Escandell, Victoria and Manuel Leonetti. 2014. “Fronting and Irony in Spanish.” In Left Sentence-Peripheries in Spanish: Diachronic, Variationist and Typological Perspectives, edited by Andreas Dufter and Octavio de Toledo, 309-342. Amsterdam and Philadelphia, PA: John Benjamins. Gallego, Ángel. 2007. “Phase Theory and Parametric Variation.” PhD diss., Universitat Autónoma de Barcelona. Green, Georgia. 1976. “Main Clause Phenomena in Subordinate Clauses.” Language 52 (2): 382-397. Haegeman, Liliane. 2000. “Negative Preposing, Negative Inversion and the Split CP.” In Negation and Polarity, edited by Laurence Horn and Yasuhiko Kato, 21-61. Oxford: Oxford UP. ––. 2002. “Anchoring to Speaker, Adverbial Clauses and the Structure of CP.” Georgetown University Working Papers in Theoretical Linguistics 2: 117-180. ––. 2006a. “Argument Fronting in English, Romance CLLD and the Left Periphery.” In Negation, Tense and Clausal Architecture: Cross-Linguistic Investigations, edited by Raffaella Zanuttini, Hector Campos, Elena Herburger and Paul Portner, 27-52. Washington: Georgetown UP. ––. 2006b. “Conditionals, Factives and the Left Periphery.” Lingua 116: 1651-1669. ––. 2007. “Operator Movement and Topicalization in Adverbial Clauses.” Folia Linguistica 18: 485-502. ––. 2010. “The Internal Syntax of Adverbial Clauses.” Lingua 120: 628-648. ––. 2012. Adverbial Clauses, Main Clause Phenomena and the Composition of the Left Periphery: The Cartography of Syntactic Structures. Oxford: Oxford UP. Haegeman, Liliane and Barbara Ürögdi. 2010. “Referential CPs and DPs: An Operator Movement Account.” Theoretical Linguistics 36 (2/3): 111-152. Hegarty, Michael. 1992. “Familiar Complements and their Complementizers: On Some Determinants of A’-Locality.” Unpublished manuscript. Hernanz, Maria Lluïsa. 2006. “Emphatic Polarity and C in Spanish.” In Studies in Spanish Syntax, edited by Laura Brugè, 105-150. Venice: Università Ca’ Foscari. Heycock, Caroline. 2006. “Embedded Root Phenomena.” In The Blackwell Companion to Syntax. Volume 2, edited by Martin Everaert and Henk van Riemsdijk, 174-209. Oxford: Blackwell. Hooper, John and Sandra Thompson. 1973. “On the Applicability of Root Transformations.” Linguistic Inquiry 4: 465-497. Jiménez-Fernández, Ángel Luis. 2010. “Discourse-Agreement Features, Phasal C and the Edge: A Minimalist Approach.” Diacrítica - Language Sciences Series 24 (1): 25-49. ––. 2011. “On the Order of Multiple Topics and Discourse-Feature Inheritance.” Dilbilim Araştırmaları [Journal of Linguistic Research] 1: 5-32.