Deconstructing the subject condition in terms of cumulative constraint violation

DOI 10.1515/tlr-2013-0022 The Linguistic Review 2014; 31(1): 73 – 150 Liliane Haegeman, Ángel L. Jiménez-Fernández* and Andrew Radford Deconstructing the Subject Condition in terms of cumulative constraint violation Abstract: Chomsky (1973) attributes the island status of nominal subjects to the Subject Condition, a constraint specific to subjects. English and Spanish are interesting languages for the comparative study of extraction from subjects, because subjects in English are predominantly preverbal, whereas in Spanish they can be either preverbal or postverbal. In this paper we argue that the islandhood of subject DPs in both English and Spanish is not categorical. The degradation associated with extraction from subjects must be attributed to the interplay of a range of more general constraints which are not specific to subjects. We argue that theinteraction of these constraints has a cumulative effect whereby the more constraints that are violated, the higher the degree of degradation that results. We also argue that some speakers have a greater tolerance for constraint violations than others, which would account for widespread inter-speaker judgment variability. Keywords: subject islands, subject positions, cumulative constraint violation, gradience, English/Spanish *Corresponding author: Ángel L. Jiménez-Fernández: University of Seville. E-mail: [email protected] Liliane Haegeman: University of Ghent. E-mail: liliane.haeg[email protected] Andrew Radford: University of Essex. E-mail: [email protected] 1 On the Subject Condition The ungrammaticality of extraction from subject nominals like those bracketed in (1) has been taken as evidence that in English subjects are islands, i.e. constituents which categorically disallow extraction.1 1 We do not deal with extraction out of clausal subjects, as in: (i) * The teacher who [that the principal would fire] was expected by the reporters is a crusty old fizzlebotch. (Ross 1986: 148, 4.251b) Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 74 Liliane Haegeman et al. (1) a. *Who did [stories about] terrify John? (Chomsky 1977: 106) b. *Who did [pictures of ] lay on the table? (Postal 1974: 189) c. *Who was [a picture of ] lying there? (Kayne 1981: 114) d. *What do you think [the joke about] offended Jack? (Pearl and Sprouse 2013: 28) e. *What did [the owner of ] sneeze? (Chaves 2013: 15) To capture the island status of subjects Chomsky (1973) postulated the Subject Condition which categorically bars extraction out of a subject phrase: (2) Subject Condition No rule can involve X, Y in the structure . . . X . . . [α . . . Y . . .] . . . where α is a subject phrase properly containing the minimal major category containing Y, and Y is subjacent to X. Chomsky’s (1973) Subject Condition was stated as a primitive constraint. With thedevelopment of the theoretical framework, the condition was subsequently reinterpreted in terms of more general constraints on extraction which were notsubject-specific but which blocked extraction out of subjects. For example, Cattell (1976) and Cinque (1977) argued that extraction is permitted only out of complements, i.e. selected constituents, and this restriction had the effect of barring extraction out of subjects, which are not selected. Likewise, Huang’s (1982a: 505) Condition on Extraction Domains/CED, permitted extraction only from a properly governed domain. Since only complements (not specifiers or adjuncts) are properly governed and subjects are specifiers within TP, CED had the effect of imposing a categorical ban on extraction out of subjects (and adjuncts).2 This is because it has been argued (e.g. by Emonds 1976, Koster 1978, Williams 1980, Stowell 1981, Safir 1986, Bresnan 1994, Postal 1998 and Alrenga 2005) that clausal subjects like that bracketed in (i) do not occupy the canonical subject position in spec-TP, but rather occupy a topic position on the edge of the clause periphery. In addition, it is unclear whether clausal subjects are CPs, or (as proposed by Takahashi 2009) DPs. 2 For a range of empirical, theoretical and experimental perspectives on CED, see Nunes and Uriagereka (2000); Sabel (2002); Rackowski and Richards (2005); Stepanov (2007); Chomsky (2008); Müller (2010); Jurka (2010); Jurka et al. (2011); Sheehan (2010, 2012); Sprouse et al. (2013). CED is potentially problematic because there are languages which allow extraction out of subjects (Stepanov 2001, 2007), or out of certain types of adjunct (Starke 2001: 40, fn.10; Truswell Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 75 In much the same way, the theory of Barriers developed in Chomsky (1986) barred extraction out of a subject in spec-TP, because extraction would involve illicitly crossing two blocking categories (corresponding to the subject DP node and the TP node), while, thanks to the possibility of VP adjunction, no such violation arose for object extraction. In more recent proposals (Uriagereka 1999), the island status of the subject in spec-TP has been related to conditions on spellout and linearization: the subject in spec-TP is a non-selected domain, whose interior is inaccessible from the outside (see Aguero-Bautista 2012: 226, Sheehan 2012 for recent summary and discussion). Other examples could be added but the net effect is that the status of the subject in spec-TP as an island for extraction is made to follow from some more general principle. The accounts cited all continue to predict that extraction from spec-TP is categorically barred.3 Since the early work, the status of the concept “Subject” itself has also changed: while the subject was originally defined in terms of one specific position in the representation of the sentence, spec-IP/spec-TP, it has since been “deconstructed” in terms of multiple subject positions (see McCloskey 1997 for a survey and motivation). Thus it became possible to account for the island status of the subject in spec-TP, a non-selected position, by invoking some general constraint on extraction while at the same time allowing extraction from subjects that occupied other (selected) positions. The approach adopted in work of this ilk has been to deliberately abstract away from other variables which make extraction from subjects easier or harder (e.g. syntactic and semantic properties of the extractee or of the matrix constituent out of which it is extracted) in the hope of gaining a deeper understanding of the question of why subjects are stronger islands than objects. However, weshall argue here that such a categorical approach obscures the true (composite) nature of extraction constraints. We shall show that, whether phrased asa subjectspecific condition or in more general terms, categorical constraints formulated to ban extraction from subjects run into problems when confronted with the empirical data: even when the subject occupies spec-TP, extraction leads to different judgments, with some cases, such as (1), being strongly degraded and others only mildly deviant. Similar variation also arises with 2007, 2009, 2011; Chaves 2012; Fábregas and Jiménez-Fernández 2012). Furthermore, sentences like (39) in the main text suggest that extraction is possible from a constituent in spec-VP, contrary to what CED predicts. 3 This is also true under the pragmatic account of subject islandhood in Erteschik-Shir (1973, 2006, 2007), if subjects are topics, and extraction never targets topics – though see Frascarelli and Jiménez-Fernández (2012, 2013) for the possibility of extracting focused constituents from what they identify as familiar topics in Spanish and Italian. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 76 Liliane Haegeman et al. subjects occupying other positions, a point that will become clearer in section 2 when we turn to Spanish data. In addition there is also speaker variation, with one speaker finding examples acceptable which for another speaker are degraded. Based on evidence from English (section 1) and Spanish (section 2), we will address these issues. With respect to extraction from subjects it will turn out that DPs in the canonical subject position in fact pattern with weak islands, i.e. structures out of which some but not all types of constituent can be extracted. Thus we must reject any prin ciple that implies a categorical ban on extraction from subjects, however formulated. We take subject islands to be non-categorical in the sense of Bianchi and Chesi (2012), and thus to be weak islands (See Bianchi and Chesi 2012 for a distinction between clear-cut rules and borderline rules; see also Bianchi and Chesi 2006 for the difference between strong and weak islands). With respect to the gradience and variability of judgments, properties of both subject and extractee will be shown to play a role in determining whether extraction is possible. In the subsections below, we explore some of the relevant factors. Our paper also illustrates how the development of the theoretical framework leads to a more refined approach to the empirical data, allowing for coarse judgments to be replaced by much more fine-grained grading of sentences. Our work is organised as follows: section 1 is a critical review of the differentfactors influencing the islandhood of DP subjects in English, taking into account both the external and the internal syntax of subjects; section 2 comparesthe English data with data from Spanish, a language which is more flexiblein the possible positions occupied by subjects. We argue that in both languages, constraints on extraction have a cumulative effect, in the sense that the more constraints that are violated, the greater the degree of degradation that results. In section 3 we argue that the different island-inducing factors identified in the data should be viewed as conditions (e.g. Inactivity Condition,Argument Condition, Intervention Condition, Specificity Condition, etc.), based on the position and properties of both the extraction site and the extractee. We note that violation of weak constraints can be alleviated by D- linking of the extracted material. In addition, we illustrate in detail the differentscenarios that emerge from the cumulative effect of constraint violations: gradience in acceptability judgments can be accounted for by proposing that the more constraints a sentence violates, the less acceptable it is. In section 4wediscuss a number of theoretical issues arising from our analysis, including the nature of the constraints we invoke, and the potential role of pragmatic and processing factors. Finally, in section 5 we summarize our main findings. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 77 1.1 The position of the subject In the literature, the island status of subjects has often been related to what we could call their “external” syntax, i.e. the position of the subject within the clause. As already mentioned, it was originally assumed that the subject nominal (the “external” argument, Williams 1994), was directly inserted (or “merged”, to use Minimalist terminology) in the canonical subject position, the specifier of IP or TP. This position is radically different from that of the complements associated with a verb, which are merged VP internally. 1.1.1 Freezing, Edge and Inactivity effects According to the VP-Internal Subject Hypothesis, developed in the mid-1980s (by Kitagawa 1986, Speas 1986, Contreras 1987, Zagona 1987, Kuroda 1988, Sportiche 1988, Koopman and Sportiche 1991, and others), subjects are first merged VP internally, as the highest argument of the verb. If subjects originate within the verb phrase and move to the canonical subject position, spec-TP, the ban on extraction out of a subject in spec-TP need no longer be stated as a primitive: extraction out of a subject in spec-TP is barred by the Freezing Principle of Wexler and Culicover (1980: 119), which for present purposes can be formulated as in (3).4 (3) Freezing Principle A moved constituent is frozen for extraction. If subjects originate within vP and move to spec-TP, it follows from (3) that no extraction will be possible out of a subject in spec-TP. Thus, the “freezing” account derives the Subject Condition. An interesting corollary of the “freezing” account is that extraction is expected to be permitted out of in situ subjects which remain in their original position within VP, but not out of ex situ subjects which move to spec-TP. Lasnik and Park (2003) argue that this claim is borne out by contrasts such as the following: 4 The Freezing Principle has been argued to be reducible to a more general locality condition (Müller 2010), or to principles of linearization and spellout (Uriagereka 1999; Nunes and Uriagereka 2000; Sheehan 2010, 2012), or labeling (Rizzi 2012), or to processing constraints (Hofmeister 2012). Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 78 Liliane Haegeman et al. (4) a. Which candidate were there [ posters of ] all over the town? b. * Which candidate were [ posters of ] all over the town? (Lasnik and Park 2003: 651) In (4a), the bracketed subject remains in its base position within VP, and extraction is possible because there is no freezing violation. By contrast, in (4b) the subject raises to spec-TP and extraction induces a freezing violation which leads to ungrammaticality.5 However, the freezing account faces a potential problem in relation to an observation due to Ross (1967), according to which passive subjects permit extraction in sentences such as (5), where the PP of which cars is seemingly extracted from the bracketed nominal in subject position: (5) Of which cars were [the hoods] damaged by the explosion? (Ross 1967: 242, 4.253) In a similar vein, Chomsky (2008) claims that wh-extraction is permitted out of passive/unaccusative subjects like that bracketed in (6a), but barred out of transitive subjects like that bracketed in (6b): (6) a. Of which car was [the driver] awarded a prize? b. *Of which car did [the driver] cause a scandal? 5 See Harwood (2012) for an alternative account of sentences like (4). A complication which we set aside here is that extraction out of a subject is barred in Predicate/Locative Inversion structures like (i/ii) and in some cases of unaccusative there clauses like (i): (i) *Which wall do you think the cause of the riot was [a picture of ]? (Moro 1997: 124) (ii) *Who do you think on this wall hung [a picture of]? (Hartmann 2005: 96) (iii) *Who did there arrive [a friend of] at the party? (Hartmann 2005: 97) See Hartmann (2005) for discussion and references. With respect to (iii), though, the data are far from clear; Radford (2009:434) gives: (iv) Of which drugs did there remain traces in the blood? See also (38a) in the main text. We leave this for future study. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 79 If the underlined wh-PP were extracted out of the subject in its superficial position in spec-TP, both sentences would be expected to induce a violation of the Freezing Principle (3) and hence to be ungrammatical.6 On the basis of contrasts like that in (6), Chomsky (2008) proposes that wh-extraction can take place successfully from the base position of the subject in sentences like (6a) – an idea dating back to work by Huang (1982b), Chomsky (1986), and Merchant (1999, 2001). If transitive subjects originate within vP and passive/unaccusative subjects originate within VP, (6a,b) will have underlying representations along the lines of (7a,b): (7) a. [CP [C ø] [TP [T was] [vP [v ø] [VP the driver of which car [V awarded] a prize]]]] b. [CP [C ø] [TP [T did] [vP the driver of which car [v ø] [VP [V cause] a scandal]]]] For Chomsky, extraction is regulated by a locality condition which can be given the following informal characterisation (where phases include CP and transitive vP): (8) Edge Condition The edge of a phase is opaque for extraction. (8) gains empirical support from the degradation in sentences like (9), where who is extracted out of a bracketed DP on the edge of a CP phase:7 (9) a. ??Who do you wonder [which picture of ] Mary bought? (Lasnik and Saito 1992: 102, 144a) b. ??Who do you wonder [which picture of ] is on sale? (Lasnik and Saito 1992: 102, 144b) The conceptual basis of the Edge Condition is that it bars “search that goes too deeply into a phase already passed” (Chomsky 2008: 148).8 In a personal communication reported in Gallego (2007: 286), Chomsky amplifies this remark by saying “Extraction from within SPEC of a phase already passed poses a locality problem, by definition. It’s necessary not only to search into the exterior of the phase 6 Our discussion here abstracts away from several complicating factors, including specificity (discussed in section 1.2.2) and pied-piping (discussed in section 1.3.1). 7 See however the discussion of (61) in section 2.2.2. 8 For discussion of apparent counterexamples to the Edge Condition and how they can be dealt with, see Gallego (2007, 2009) and Boeckx (2012: 131–132). Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 80 Liliane Haegeman et al. already passed (which is clearly OK), but also one level of depth further, into the interior of that exterior.” The Edge Condition receives independent support from experimental research by Jurka (2010) on German. Using a 7-point Likert scale onwhich 7 represents the highest and 1 the lowest level of acceptability, Jurka reported that extraction out of an in situ transitive subject in spec-vP yields a substantially lower score of 3.55 than the score of 6.14 for extraction out of an in situ object in the complement position of the verb (abbreviated for convenience as ‘comp-VP’). If vPs without an external argument like that in (7a) are not phases, nothing will prevent the PP of which car from being extracted out of the underlined subject and moving to spec-CP. But if vPs with an external argument arephases, extracting of which car out of the underlined subject and moving ittospec-CP in (7b)will be ruled out by Edge Condition (8) because the underlined subject is the specifier of a vP phase. Extraction out of the subject in its superficial position in spec-TP will seemingly be ruled out by the Freezing Principle (3). However, the freezing account runs into potential problems in relation to examples such as the following (from Chomsky 2008, ex. 19, and treated as fully grammatical by him): (10) a. Of which car is [the driver] likely to cause a scandal? b. Of which car did they believe [the driver] to have caused a scandal? On Chomsky’s assumptions, the DP the driver of which car originates as the specifier/subject of the embedded (phasal) vP, and then moves through an intermediate position on the edge of the TP headed by infinitival to before reaching its superficial position in the matrix clause in spec-TP as the subject of is/was in (10a), and in spec-VP as the object of believe in (10b), with the verb believe raising to adjoin to the head v of vP. In its base position in spec-vP, extraction out of the bracketed DP is barred by the Edge Condition, and in its intermediate and superficial positions it is blocked by the Freezing Condition. Thus, the account sketched above wrongly predicts that sentences like (10) are ungrammatical. To circumvent this problem, we have to abandon or modify one of the existing constraints. If we abandon the Edge Condition, we are seemingly left with no account of the contrast in (6). This suggests that we need to replace the Freezing Principle by a more targeted constraint. But what? Chomsky (2008: 150) suggests that the answer lies in an Inactivity Conditionwhich makes an A-chain invisible to further computation (and hence opaque to extraction) once it has been rendered inactive by the valuation of its uninter- Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 81 pretable features. For present purposes, we can formulate this condition as follows:9 (11) Inactivity Condition An inactive A-chain (i.e. one with no unvalued A-features) is opaque for extraction We can illustrate the effect of (11) by considering the derivation of (10a). The DP the driver of which car will originate in spec-vP as in: (12) [CP [C ø] [TP [T is] likely [TP [T to] [vP the driver of which car [v cause] a scandal]]]] The Edge Condition (7) prevents C from probing at this point to attract the PP of which car to move to the edge of CP. Instead, T-to probes and attracts DP the driver of which car to become its specifier, so deriving (13): (13) [CP [C ø] [TP [T is] likely [TP the driver of which car [T to] [vP t [v cause] a scandal]]]] If T-is probes before C in (13), T-is will agree with and assign nominative case to DP the driver of which car and attract it to move to spec-TP, deriving (14): (14) [CP [C ø] [TP the driver of which car [T is] likely [TP t [T to] [vP t [v cause] a scandal]]]] The Inactivity Condition (11) will then prevent C from probing in (14) to attract thePP of which car to move to spec-CP, because the DP the driver of which car is inactive by virtue of its case feature having been valued, and so is opaque for extraction. So, at first sight, it might seem as if Chomsky’s analysis wrongly predicts that sentences like (10a) should be ungrammatical. However, this is not the case, since there is an alternative derivation for (10a)which does not induce an inactivity violation. To see this, let’s return to the stage of derivation in (14). This time, let us suppose that C probes before T-is. If so, the Inactivity Condition will not bar C from attracting PP of which car to moveto spec-CP (since the uninterpretable case feature on the subject DP will be 9 The Inactivity Condition is arguably a subcase of the Freezing Principle which is restricted to A-chains. See Richards (2011) for an argument against the Inactivity Condition, and Boeckx (2012: 105–106) for a rebuttal and for a defence of the conceptual rationale of the condition. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 88 Liliane Haegeman et al. peripheral Focus Phrase projection. More importantly for our analysis of English, the relevant constraint does not hold for English, since DP-internal PPs can bemodified by a focus particle, as we see from the following internet-sourced examples: (28) a. They have no trace of a circulation, and [traces only of nerves ] ( Journal of Practical Medicine 19: 218) b. [Traces only of resin , gum and extractive matter] can be separated from the mass . . . (Philosophical Magazine: 17) c. . . . we find [the remains only of marine plants and animals ] (Brewster’s Edinburgh Encyclopaedia vol. 13: 435) Nor is Jurka’s experimental evidence any more compelling. Indeed, he reports a “statistically significant” difference between extracting a PP out of a DP subject in English (score = 3.29) and extracting a PP out of a DP object (score = 3.86). If both PPs are generated in situ, this effect remains unaccounted for. By contrast, if the PP is extracted from its containing DP, we expect the asymmetry because extraction from a moved subject should yield a freezing violation which does not hold in the case of extraction from an object. The somewhat low score of 3.86 for extracting a PP out of an object can be accounted for by the perceived unnaturalness of pied-piping in colloquial English in contexts where preposition stranding is also possible. The very low score of 2.51 for extraction of a DP out of a subject can be accounted for by the fact that this involves violation of an additional constraint on preposition stranding discussed in Section 1.3.1. Overall, then, we conclude that there is no compelling evidence in support of a categorical ban on extracting PPs out of DPs. This conclusion will turn out to be particularly important for the discussion of Spanish, where all cases of extraction that we will discuss involve fronting of PP rather than of DP. 1.1.4 On extraction from non-terminal positions One of the most interesting theoretical aspects of Chomsky’s analysis is his claim that for moved subjects extraction is possible from their non-terminal positions (e.g. in their base or intermediate positions). However, this claim is potentially problematic from a conceptual point of view for two reasons. Firstly, allowing extraction from the base, intermediate or superficial position of the subject in sentences like (10) means that the ultimate acceptability status of a given sentence can only be determined by a transderivational comparison of a number of Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 89 alternative competing derivations of the same sentence (e.g. one where extraction takes place from the base position, another where it takes place from the intermediate position, and a third where it takes place from the superficial position), andsuch an assumption leads to an undesirable increase in the power and complexity of grammars. Secondly, extracting from the base rather than the superficial position of a moved subject leads to an economy violation, in that it results inthe formation of a longer wh-chain than which would arise if extraction took place from the superficial position of the subject.12 Furthermore, the empirical robustness of the data on which Chomsky’s analysis is based is in dispute. For example, contrary to his claim that extraction from ECM subjects is fine, extraction from ECM subjects in sentences like (29) has been argued to lead to ungrammaticality: (29) a. * Of whom does Mary believe [ friends] to be stupid? (Sabel 2002: 293) b. * Who do you expect [stories about] to terrify John? (Chomsky 1973) c. * Which artists did you find [works by] to be offensive? (Uriagereka 2004: 10) d. * Who did John believe [pictures of] to have caused the riot? (Boeckx 2012: 116) Moreover, Broekhuis (2006) presents evidence from Dutch sentences like (30) against Chomsky’s assumption that extraction is possible from the base position of a moved subject: (30) a. Wat zijn (er) jouw vader [voor rare verhalen] verteld? what are-3pl (there) your father for strange stories told ‘What kind of strange stories were (there) told to your father?’ b. *Wat zijn [voor rare verhalen] jouw vader verteld? what are-3pl for strange stories your father told ‘What kind of strange stories were told to your father’ 12 In addition, we note that extraction from the intermediate position of the subject (as the specifier of infinitival to) in structures like (14) is potentially problematic in terms of Chomsky’s own assumptions. If A-movement is contingent on agreement (e.g. if T attracts the closest constituent it agrees with in one or more φ-features), and if T inherits its agreement features from C, it is not clear how the subject can move to spec-TP if (as generally assumed) ECM and raising infinitives are defective clauses which project TP but not CP, since T will have no C to inherit agreement from. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 90 Liliane Haegeman et al. Extraction is possible from the base position of the in situ subject in comp-VP bracketed in (30a), but not from the superficial position of the ex situ subject inspec-TP in (30b). This finding is unexpected under Chomsky’s analysis, since extraction should be possible from the base position of the subject in (30b), andwould thus be expected to yield an outcome which is just as grammatical as(30a). Experimental research by Jurka (2010) further undermines Chomsky’s analysis. In an experiment on German (where transitive subjects can occupy spec-vP, spec-TP or spec-CP), Jurka found (2010: 63, Table 3.2) that the mean acceptability score (on a 7-point scale) for extraction out of an in situ transitive subject in spec-vP was 3.55, whereas it was significantly lower (2.28) for extraction out of a subject in spec-TP. If extraction from a transitive subject which has moved to spec- TP were possible from its base position in spec-vP, the acceptability of extraction out of a spec-TP subject should be on a par with extraction from a spec-vP subject, whereas in actual fact it is substantially lower. Moreover, Jurka also reports that extraction from in situ objects in German yielded a markedly higher acceptability score (6.14) than extraction from a moved object (2.84). These findings suggest that extraction is always from the superficial position of a constituent. This generalisation can be captured in terms of the following constraint: (31) Extraction Constraint Extraction is only possible from the head of a chain. This is not a primitive constraint per se, but rather follows from more general locality principles such as Chomsky’s (1995) Minimal Link Condition/MLC (requiring movement to be as local as possible), with MLC itself being a reflex of a more general Economy condition. Jurka also conducted a series of experiments on English which further undermine the empirical foundation of Chomsky’s analysis. Recall that Chomsky claims that extraction is possible out of subjects which undergo A-movement in sentences like (10) (e.g. ECM subjects); he also claims that extraction is possible from unaccusative and passive subjects which undergo local A-movement in sentences like (6a), but not from their transitive counterparts like (6b). However, Jurka’s findings cast doubt on the observational adequacy of these claims. In one experiment, Jurka found that extraction from ECM subjects (e.g. in Which politician did John believe a book about to have caused a scandal?) received a low acceptability score of 2.24, which was even lower than the 2.61 score for extraction out of a local transitive subject (e.g. in Which politician did a book about cause a scandal?), thereby calling into question Chomsky’s claim that extraction from ECM subjects is acceptable. In a second experiment, Jurka found that extraction Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 91 from a passive subject (e.g. in John wondered which man a book about was released last year) yielded a score of 2.68, which was not significantly different from the score of 2.55 for extraction out of the corresponding active subject ( John wondered which man a book about caused a scandal last year). These findings provide experimental support for the Extraction Constraint (31), and strongly suggest that the base and intermediate positions of a moved subject do not affect its extractability. This in turn means that Chomsky’s analysis needs to be modified in the light of experimental support for the Extraction Constraint. But how? In order to answer this question, consider how to account Jurka’s findings for was . . . für ‘what . . . for’ extraction in German. On one experiment (2010: 88) he reports acceptability scores of 4.74/4.65 for extraction out of in situ passive/unaccusative subjects, but a significantly lower score of 3.51 for extraction out of in situ transitive agentive subjects. On another (2010: 63), he reports scores of 3.55 for extraction out of an in situ transitive agentive subject in spec-vP, but a significantly lower score of 2.28 for extraction out of a transitive agentive subject in spec-TP. Extraction out ofan in situ passive/unaccusative subject in comp-VP will not induce an Edge violation, nor a Freezing violation, nor an Inactivity violation (as long as C probes before T values the case feature on the subject), nor an Extraction violation (since extraction takes place from the head of a trivial A-chain). Extraction from an in situ agent subject in spec-vP will be degraded because it violates the Edge Condition, but not the Freezing, Inactivity or Extraction conditions. Extraction from an ex situ subject in spec-TP will be doubly degraded because it violates both the Inactivity Condition (since a subject in spec-TP will be inactive by virtue of having had its case feature already valued as nominative) and the Freezing Condition. Now consider Jurka’s findings for English, where extraction of a PP from a passive theme subject yielded a very low score of 2.68, which was not significantly different from the score of 2.55 for extraction out of the corresponding transitive agent subject. These very low scores can be accounted for by positing that (in consequence of the Extraction Condition) extraction in both cases takes place from the superficial position of the subject in spec-TP and involves a double constraint violation, since both the Freezing Principle and the Inactivity Condition are flouted. A similar violation of the same two constraints will occur in cases of extraction from an ECM subject. The somewhat lower score of 2.24 for extraction out of an ECM subject can be attributed to the additional complexity (and perhaps infrequency) of ECM structures, given that Jurka (201: 161) reports that ECM structures in non-extraction contexts receive a much lower acceptability score than their non-ECM counterparts (e.g. John believed a book about Obama to have caused a scandal received a score of 4.48 compared to 6.77 for A book about Obama caused a scandal ). Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 92 Liliane Haegeman et al. To summarise: This section has been concerned with how the position of asubject affects the possibility of extracting out of it. Chomsky (2008) argued thatextraction can (in principle) take place from the base, intermediate or superficial position of a subject, but noted that particular types of extraction can induce constraint violations which cause degradation. More specifically, Chomsky claimed that extraction from subjects which undergo local A-movement to the specifier position of a finite TP is barred from their superficial position by the Inactivity Condition, and barred from their base position (for subjects originating in spec-vP) by the Edge Condition: by contrast, he posited that extraction from subjects which undergo long-distance movement is possible fromintermediate positions. However, we saw that research by Jurka (2010) provided experimental evidence that extraction is only possible out of the head of a chain (an effect which we captured in terms of the Extraction Constraint/ EC 31). We offered an alternative account of extraction from subjects under which(in consequence of EC), extraction is only possible from the head of a chain, with the Edge Constraint yielding degradation when extracting out of an insitu agent subject on the edge of a vP phase, and the Inactivity Condition andthe Freezing Principle yielding double degradation when extracting out of asubject in the specifier position of a finite TP (or, in the case of ECM subjects, inspec-VP). 1.2 Properties of the subject In addition to being affected by the external syntax of the subject, i.e. its position in the clause, the possibility of extraction out of subjects is also affected by the internal properties of the subject itself. We have already discussed the issue of whether thematic properties of subjects affect extraction from them in Sections 1.1.2 and 1.1.4. In this section, we look at the impact of specificity on extraction possibilities. Among others, Horn (1974), Hornstein (1977), May (1977), Chomsky (1977, 1981), Cinque (1990), Mahajan (1992), Ormazábal (1992), Chung (1994), Kluender (1998, 2004), Stepanov (2001), Davies and Dubinsky (2003), and Goodall (2004) have observed that extraction is barred out of a specific nominal introduced e.g.by a demonstrative like that or a possessive like your, but not out of a nonspecific nominal introduced e.g. by the indefinite article a or a quantifier like several. The following examples show that extraction is more readily permitted out of non-specific nominals like the bracketed object in (32a) or the bracketed subject in (32c) than out of specific nominals like those bracketed in (32b) and (32d): Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 93 (32) a. Of what did he want [a picture ---]? b. ?Of what did he want [that picture ---]? c. ??Of what did [pictures ---] upset him? d. *Of what did [those pictures ---] upset him? Fiengo and Higginbotham (1981) propose a Specificity Condition which can be formulated within the spirit of their proposal as follows: (33) Specificity Condition Specific nominals are opaque domains for extraction.13 As should be obvious, the Specificity Condition is again not specific to subjects, but holds of all DPs: specificity of the DP makes extraction more difficult. A number of different accounts have been proposed of the Specificity Condition, but for succinctness we will mention only one here. Campbell (1996) posits that specific/referential nominals contain an abstract specificity operator in spec-DP which blocks extraction. On this view, the DP bracketed in (32d)would have the structure (34) below, where OP is an abstract specificity operator:14 (34) [DP OP [D those] pictures of what ] Extraction of what out of its containing DP in (34) results in a structure in which the wh-operator what crosses the intervening specificity operator OP, thereby incurring a violation of the Intervention Constraint discussed in section 1.3.3.15 13 See Enç (1991) for arguments that specificity rather than definiteness is the factor responsible for creating opaque domains. Experimental evidence in support of a specificity effect comes from Sprouse and Almeida (2012a). We abstract away here from other factors affecting extractability, including the semantic properties of the associated predicate (Diesing 1992; Erteschik-Sher 1981; Kluender 1992; Keller 2000). 14 (34) is simplified, inter alia, by not showing the launch site of the operator which moves to spec-DP: see Campbell (1996) and Aboh (2004) for evidence of movement internally within DP. Haegeman and Ürögdi (2010a,b) extend the null specificity operator analysis to factive complement clauses and claim that the operator is what renders the clause “referential”. The assumption that an Operator in spec-DP makes DP a barrier to extraction raises interesting questions about why DP loses its barrierhood when its head is extracted (Stepanov 2012). 15 See Baunaz (2012: 37–38) for a more fine-grained analysis of DPs in which three degrees of extractability are distinguished. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 94 Liliane Haegeman et al. 1.3 Position and properties of the extractee In addition to being sensitive to the position (external syntax) and properties (internal syntax) of the constituent out of which it takes place, extraction is also sensitive to the external syntax and the internal syntax of the extractee itself. 1.3.1 Pied piping In relation to the external syntax of the extractee, consider contrasts such as the following (noted in Ross 1967, 1986; Chomsky 1977, 1986 and much subsequent work): (35) a. He is the only player of whom [ pictures] were taken. b. * He is the only player who [ pictures of ] were taken. Why is the preposition piedpiping structure (35a) more acceptable than the preposition stranding example in (35b)? This is all the more puzzling as in general preposition stranding is found to be more acceptable than pied-piping in English. For example, Jurka (2010: 154) reports the results of an experiment which showed that stranding the preposition under extraction from an object in structures like (36a) received a far higher acceptability score (parenthesized) than pied-piping the preposition: (36) a. Phil wondered which topic Scott had filmed [a documentary about] last year. (5.08) b. Phil wondered about which topic Scott had filmed [a documentary] last year. (3.86) The contrast in (36) follows if (as suggested by Chomsky 1995: 262) pied-piping only takes place when required for convergence (e.g. when some constraint bars stranding).16 This line of reasoning suggests that the preposition of is pied-piped along with who(m) in (35a) because some constraint prevents the preposition from being stranded in (35b). But what is this constraint? Kuno (1973) attributes the ill-formedness of sentences like (35b) to violation of the following constraint: 16 An interesting question which we set aside here is why (as noted by Sag 2010) pied-piping is generally more acceptable in relative than in interrogative clauses. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 95 (37) Incomplete Subject Constraint It is not possible to move any element of a subject noun phrase/clause if what is left over constitutes an incomplete noun phrase/clause. (Kuno 1973: 380) Kuno defines incompleteness as follows: “A noun phrase/clause is incomplete if an obligatory element is missing. Thus, the [NP Prep] pattern is incomplete because the object of the preposition is missing” (Kuno 1973: 380). (37) is intended to capture the generalization that a preposition can’t be stranded inside a subject in a sentence like (35b). The theoretical status of (37) is unclear and it is also empirically inadequate. As the examples below illustrate, a preposition can be stranded inside an in situ subject as in (38a), thus violating (37), but not inside an ex situ subject as in (38b): (38) a. There are several players whose fitness there remain [doubts about]. b. * There are several players whose fitness [doubts about] remain. At first sight, it might seem as if the contrast in (38) could be handled in terms of the linear position of the preposition in relation to the clause. It could be argued that stranded preposition must not be followed by any clausal material and must thus effectively occur on the right edge of the clause. However, any such constraint would be called into question by attested (internet-sourced) examples like (39), all of which contain a preposition stranded clause-internally and followed by additional material: (39) a. . . . and has already cut her first CD, which I can send [samples of ] to anyone that’s interested. b. Who can I talk [to] about my depression? c. What do astronauts like to take [pictures of ] from space? d. You have the stress of (as yet) unresolved debt payments which you will need to get [advice on] from debt counsellors. e. There is the Access to Learning Fund available on a criteria basis which you can get [information about] from the University. Instead, it seems more plausible to take the contrast in (38) to be an effect of the Freezing Principle (3), whereby a preposition can be stranded inside an in situ but not an ex situ (i.e. moved) constituent. Pursuing this possibility, let us posit the following constraint: Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 96 Liliane Haegeman et al. (40) Preposition Stranding Constraint/PSC A preposition cannot be stranded inside a moved constituent. Evidence in support of (40) comes from contrasts such as the following: (41) a. Tell me who you’re touching up [a picture of ]? b. ??Tell me who you’re touching [a picture of ] up? (Kayne 2002: 74) (42) a. Who do you think that John wanted [ pictures of ]? b. ?*Who do you think that [ pictures of ] John wanted? (Stepanov 2007: 102, 41b) In (41a/42a), the bracketed direct object remains in situ, and the preposition of/ about can be stranded inside it. By contrast, in (41b/42b) the direct object moves – to a position above the particle up in (41b) and to a position above the subject John in (42b) – and stranding a preposition inside an ex situ object leads to a much lower level of acceptability. In the light of PSC (40), consider the relative acceptability scores (indicated in parentheses, using a 7-point scale) reported by Jurka (2010) for sentences such as (43): (43) a. Phil wondered about which topic [a documentary] had swayed the voters last year. (3.29) b. Phil wondered which politician [a documentary about] had swayed the voters. (2.51) Given the Extraction Condition (31), extraction will take place out of the superficial (spec-TP) position of the subject in both examples. This means that both incur a Freezing violation (since the subject has moved from its initial position within vP) and also an Inactivity violation (since once it reaches spec-TP the subject has had its case feature valued and is therefore inactive). However, (43b) will additionally incur a violation of the Preposition Stranding Constraint (40), since the preposition about has been stranded inside a moved constituent (i.e. inside a subject which has moved to spec-TP).17 17 A factor which we set aside here is that, as noted by Chaves (2013), P-stranding violations can be ameliorated by the presence of a parasitic gap, e.g. in (i): (i) Who did [the rivals of ---] shoot ---? (Chaves 2013: 7, 8a) Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 97 An interesting side-issue which we note in passing is that although sentences like (35a) involving extraction of a PP out of a subject in spec-TP are implicitly treated as fully acceptable in much of the syntactic literature, studies like Jurka’s show that they are actually given quite low acceptability ratings under experimental conditions (e.g. 3.29 on a 7-point scale for 43a). This low score is consistent with our earlier claim that they incur Freezing and Inactivity violations. 1.3.2 Argument vs. adjunct Consider (44): (44) a. ?* Of which dress did [the designer] cause a scandal? b. * From which fashion house did [the designer] cause a scandal? In both examples extraction takes place from a specific, inactive, moved DP at the head of an A-chain. Hence both (44a) and (44b) lead to Specificity, Inactivity and Freezing violations. If these conditions were absolute then they would wrongly predict both examples to be equally ill-formed, contrary to fact (see Starke 2001: 34–5 for a similar contrast from French). However, it has long been known that the semantic properties of the extractee play a role in regulating extraction. The relevant factors claimed to play a part in determining extractability include theta roles (Rizzi 1990), case (Manzini 1992), individuation (Frampton 1991; Cresti 1995) and richness of internal semantic structure (Szabolcsi and Zwarts 1997). For instance, Rizzi (1990: 86) observed that only constituents with a “referential” thetarole (e.g. one like agent, theme, goal referring to a participant in the event described by the verb) can be extracted out of weak islands. The effect of this constraint can be illustrated by the following examples involving extraction of an (underlined) constituent out of a (bracketed) weak wh-island: (45) a. * How did he wonder [whether to fix the car]? b. ??What did he wonder [whether to fix]? c. ?This is the car which he wondered [whether to fix]. d. ?Which car did he wonder [whether to fix]? More generally, a structure where a moved constituent is associated with an illicit gap inside an island is ameliorated if the moved constituent is also associated with a licit gap not inside an island: see Phillips (2006) for experimental evidence. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 104 Liliane Haegeman et al. examples below, it differs from English in banning prepositions from being stranded anywhere, even inside an in situ complement PP as in (53b), and requires pied-piping of the preposition instead, as in (53a): (53) a. ¿Con quién hablaba Juan? with whom was.speaking-3sg Juan? ‘With whom was Juan speaking?’ b. *¿Quién hablaba Juan con? whom was.speaking-3sg Juan with? ‘Who was Juan speaking with?’ The key difference between Spanish and English appears to be that Spanish bans a preposition from being stranded in any position, whereas English bans a preposition from being stranded inside a moved constituent. This suggests that the Preposition Stranding Constraint may be parameterised in the manner specified informally below, where the % diacritic means that the parenthesized condition holds in some languages (e.g. English), but not in others (e.g. Spanish). (54) Preposition Stranding Condition/PSC (revised; final formulation) No preposition can be stranded (% inside a moved constituent). A further parametric difference is that PSC appears to be a much stronger constraint in Spanish than in English.22 Thus, Chaves (2013: 13) maintains that the PSC violation that arises from stranding a preposition inside a moved subject in spec-TP in sentences such as (55) can be alleviated with the prosodic phrasing marked by the square brackets, because this cues where the extraction occurs: (55) a. [Which doctors] [have patients of ] [ filed malpractice suits in the last year]? b. [Which problem] [will a solution to] [never be found]? By contrast, PSC violations in Spanish lead to irreparable ungrammaticality (i.e.ungrammaticality which cannot be alleviated e.g. by prosodic phrasing or D-linking). The global ban on preposition stranding in Spanish means that a prepositional complement can only be extracted out of a subject in Spanish if the preposition is pied-piped along with its complement, not if the preposition is stranded. 22 Ideally we would like to relate this parameterization to other properties of the languages, but this must await future research. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 105 We illustrate this in (56), where extraction takes place from a postverbal subject which, as we will see presently, is more accessible to extraction than a preverbal subject. As illustrated in (56a), a prepositional complement cannot be extracted on its own from within a subject DP, even if the subject is postverbal. Instead, the whole PP has to be extracted, as in (56b): (56) a. *¿Qué príncipe fueron publicadas [varias fotos which prince were-3pl published several photos comprometedoras de]? compromising of? ‘Which prince were several compromising photos of published?’ b. ¿De qué príncipe fueron publicadas [varias fotos of which prince were-3pl published several photos comprometedoras]? compromising ‘Of which prince were several compromising photos published?’ We will take this property of Spanish for granted in subsequent discussion, so that all examples involving extraction of a prepositional object out of a subject will illustrate fronting of the containing PP. 2.2 External syntax 2.2.1 Preverbal subjects vs. postverbal subjects Based on (57), taken from Martí (1999)23, Starke concludes (2001: 57) that Spanish is a type II language; preverbal subjects like that bracketed in (57a) are opaque for extraction, whereas postverbal subjects like that bracketed in (57b) are transparent. 23 We note, however, that (contrary to Martí’s judgments) most of our informants did not consider (57a) markedly worse than (57b), so it would seem that for more liberal speakers, the constraint against extraction from a (non-specific) preverbal subject in Spanish is a mild one. We will attempt to account for inter-speaker judgment variability in section 3. It may be that the whphrase is D-linked in (57), ameliorating extraction. The interrogative quantifier qué in Spanish corresponds to either ‘what’ or ‘which’ in English, with the discourse setting determining whether it is D-linked. Spanish also has the interrogative quantifier cuál, which is unambigu ously D-linked and so always corresponds to English ‘which’, but is used much less frequently than qué. We will assume that in the examples (57) qué is D-linked and hence corresponds to English which. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 106 Liliane Haegeman et al. (57) a. ??*¿De qué autor crees que [varios libros] han of which author believe-2sg that several books have-3pl recibido premios internacionales? received awards international-pl b. ?¿De qué autor crees que han recibido premios of which author believe-2sg that have-3pl received awards internacionales [varios libros]? international various books ‘By which author do you think several books have received several international awards?’ (Starke 2001: 57, ex. 135 a–b) Adopting Uribe-Etxebarria’s (1992) claim that Spanish subjects are generated in spec-vP and are subsequently moved to spec-TP, Martí (1999) holds, after Takahashi (1994), that extraction from a subject DP in spec-TP in a sentence like (57a) violates the Subject Condition (2), which is taken to be a consequence of the Freezing Principle (3) discussed earlier. By contrast, she assumes that postverbal subjects like that bracketed in (57b) remain in spec-vP, which is not a freezing position, and hence in situ subjects allow extraction. Martí (1999) concludes that extraction is allowed only if the subject remains in situ in spec-vP. As should be obvious, her account is consistent with the conclusion we drew in section 1.1.4 that extraction always takes place from the superficial position of a subject, in consequence of the Extraction Constraint (31). If in situ goal subjects occupy a position below the phase edge (as we suggested in section 1.1.2), there will be no violation of the Edge Condition (8). Observe, though, that Martí judges (57b) to be slightly marginal, which is unexpected on her account. 2.2.2 Phases and phase sliding Adopting the Minimalist framework in general and Phase theory in particular, Gallego and Uriagereka (2007) posit that extraction from a constituent on a phase edge is barred by the Edge Condition (8). More specifically, phases are uniform cross-linguistically, and CP and v*P (i.e. a vP with an external argument) are phases. Thus constituents on the edge of CP and v*P disallow extraction. However, in Gallego and Uriagereka’s approach, phasehood is not immutable: a phase may lose its status as a phase and as a consequence its edge will allow extraction. When v*P loses its phasehood, extraction from the edge of v*P becomes licit. In Spanish, v-to-T movement results in Phase Sliding, by which TP inherits phasehood from v* (See den Dikken 2006, 2007 for a similar notion of Phase Extension). Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 107 As a result of Phase Sliding, v*P is no longer phasal and TP becomes the phase. Consequently, the Edge Condition (8) will bar extraction from a preverbal subject on the edge of TP, but it will no longer bar extraction from a postverbal subject on the edge of vP. Accordingly, based on Spanish data originally from Uriagereka (1988), extraction is banned in (58b) because the subject is in spec-TP, whereas it is permitted in (58a) because the subject is in situ in the specifier position of a vP which has lost phasehood as a result of the verb impresionar raising to some functional head position above vP. If TP is a phase in Spanish, the Edge Condition (8) correctly bars extraction out of the preverbal bracketed subject DP in (58b), since the subject is a specifier on the edge of a TP phase.24 (58) a. ¿De qué conferenciantes i te parece que mez of which speakers cl-2sg appear-3sg that cl-1sg van a impresionarv [v*P [DP las propuestas ti ] tz tv ]? go-3pl to impress the proposals b. *¿De qué conferenciantesi te parece que [DP las of which speakers cl-2sg appear-3sg that the propuestas ti ]j mez van a impresionarv [v*P tj tz tv ]? proposals cl-1sg go-3pl to impress ‘Which speakers does it appear to you that the proposals by will impress me?’ However, Jiménez-Fernández (2009, 2012) has argued at length against Gallego and Uriagereka’s analysis. One of the main empirical shortcomings of their analysis is that their claim that the Spanish TP is a phase and that vP is not wrongly predicts that, at least for transitive verbs, extraction will be permitted from postverbal but not from preverbal subjects. The data in (58) are in line with this prediction but the data are far from uniform. In particular, where the subject is non-specific, extraction is permitted from both pre- and post-verbal subjects, as shown in (59): 24 As will be apparent, the grammaticality judgments given by Uriagereka differ from those of Martí. For Martí, extraction from a subject always induces some sort of degradation, whereas for Uriagereka degradation arises only when the subject is preverbal, hence in spec-TP. A confounding factor is that extraction takes place out of a “specific” subject in (58), incurring an additional (specificity) violation – and this may well be why most of our informants found both examples in (58) degraded. Such divergences in judgments raise the general problem of dealing with interspeaker variability in judgments. We return to this issue in Section 3. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 108 Liliane Haegeman et al. (59) a. ¿De qué cantante te parece que me van a of what singer cl-2sg appear-3sg that cl-1sg go-3pl to escandalizar [varias fotos]? shock several photos b. ?¿De qué cantante te parece que [varias fotos] of what singer cl-2sg appear-3sg that several photos me van a escandalizar? cl-1sg go-3pl to shock ‘Of which singer does it appear to you that several photos will shock me?’ If TP were a phase, extraction out of the preverbal subject in spec-TP in (59b) should be categorically barred by the Edge Condition (8). However, as shown bythe contrast between (58b) and (59b), extraction out of a subject in spec-TP is not uniformly bad. In both examples the extraction targets the subject in spec-TP: (58b) is ungrammatical, but (59b) is much better. The difference between the two examples is that in (58b) the subject las propuestas de qué conferenciantes is specific, whereas that in (59b), varias fotos de qué cantante is not. Thespecificity of a constituent makes extraction from it harder. The degradation detected in (59b) is due to violation of the Freezing Condition (incurred by extracting PP out of a moved subject) and of the Inactivity Condition (incurred by extracting PP out of a subject which is inactive by virtue of having had its case feature valued). A further problem posed by Gallego and Uriagereka’s analysis is that it wouldseem to make incorrect predictions about extraction out of subjects of monoargumental clauses with unaccusative, passive and raising predicates. If(as widely assumed), vP is not a phase in such clauses25, it follows that even inthe context of v-to-T movement, there will be no phase sliding and TP cannotinherit phasehood from vP. If so, the prediction is that extraction should freely be per mitted out of either preverbal or postverbal subjects in intransitiveclauses. However, this prediction is not borne out by the following (passive) examples: (60) a. ?¿De qué coches parece que fueron arrestados of which cars seem-3sg that were-3pl arrested [los conductores]? the drivers 25 However, see Legate (2003) for a contrary view. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 109 b. ??¿De qué coches parece que [los conductores] fueron of which cars seem-3sg that the drivers were-3pl arrestados? arrested ‘Of which cars does it seem that the drivers were arrested?’ In these two sentences the subject of the passive predicate arrestados ‘arrested’ originates as complement of VP. In (60a) the subject remains in situ and extraction of the PP takes place from this position; since a passive vP is not a phase, there is no phase-based violation and yet the sentence is degraded. In (60b) extraction takes place from spec-TP, which in this case would not become a phasal edge through phase sliding, predicting full acceptability. However, example (60b) is even more degraded than (60a). In other words, regardless of the absence of phase sliding, extraction from spec-TP may yield a degraded outcome, which indicates that one or more other constraints have been violated. (In our terms, (60a) and (60b) both violate the Specificity Condition, and (60b) also violates the Inactivity and Freezing conditions.) A final problem for Gallego and Uriagereka’s analysis is that their proposal is dependent on the viability of the Edge Condition. The descriptive adequacy of the Edge Condition for languages like Spanish and Italian has been called into question on the basis of examples like (61), where one wh-phrase seems to have been extracted from within another wh-phrase located on the edge of a CP phase, in violation of the Edge Condition.26 Chomsky (1986: 26, 49b) provides Spanish (61a) attributed to Esther Torrego, and Rizzi (2006: 114) gives Italian (61b): (61) a. ¿De qué autora no sabes [qué traducciones] han of what author-f not know-2sg what translations have-3pl ganado premios internacionales? won awards international-pl ‘By which author don’t you know what translations have won international awards?’ b. Di quale autore ti domandi [quanti libri] siano of which author cl-2sg wonder-2sg how.many books are-3pl stati censurati? been censored ‘By which author do you wonder how many books have been censored?’ 26 For further discussion of similar structures, see Rochemont and Culicover (1990), Lasnik and Saito (1992), and Maeda (2010). Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 110 Liliane Haegeman et al. However, it is not clear that such examples are legitimate counterexamples. Observe that the wh-constituent at stake here is a subject. One option would be toassume that in both examples the wh-subject is in fact not moved to the left periphery at all, but remains in its clausal subject position (cf. Agbayani 2000, Chomsky 2013). On this assumption, extraction would not be from the edge of CP but from a subject in spec-TP.27 Another strategy, pursued in Gallego (2007: 340), is to argue that in sentences like (61) ‘the alleged sub-extracted PP is actually base generated outside the embedded wh-phrase, as a PP dependent of the matrix verb: an aboutness phrase’. This hypothesis is argued for at length by Gallego (2007: 335–354) and additional support is provided by Boeckx (2012: 131–132). So, on this account too, sentences like (61) do not undermine the Edge Condition. We will therefore not pursue these data here. Gallego (2011) goes back on the earlier analysis of Gallego and Uriagereka (2006, 2007); to account for the contrast in (58), he draws a distinction between freezing and non-freezing positions, and follows a well-established tradition (Ormazábal et al. 1994; Takahashi 1994; Boeckx 2003; Rizzi 2006; Stepanov 2007; Chomsky 2008) in claiming that spec-TP is a freezing position, but spec-vP is not. For him extraction from the postverbal subject in (58a) is licit in Spanish because the subject remains in spec-vP, a non-freezing position. Conversely, if the subject moves to spec-TP extraction is banned, as in (58b), because spec-TP is a freezing position. However, Gallego’s analysis faces the empirical problem that it wrongly predicts that sentences like (58b) are categorically ungrammatical in Spanish, whereas in fact the pattern is more complex: such extractions are mildly degraded out of a non-specific subject as in (59b) but they become more severely degraded when extraction is out of a “specific” subject as in (58b). From the data discussed above it emerges that, as was the case for English, inSpanish extraction from a subject DP gives rise to varying degrees of acceptability. This is due to the interplay of different conditions. Extraction from a preverbal subject in spec-TP will systematically yield a violation of the Freezing Principle (3) and the Inactivity Condition (11), and will lead to additional degradation if the subject is specific, since there will be concomitant violation of the Specificity Condition (33). By contrast, extraction from a postverbal subject in a sentence like (59a) is relatively acceptable. Under our analysis, this is because there is noviolation of the Freezing Condition (because the subject remains in situ), nor of the Specificity Condition (because the subject is non-specific) nor of the Edge Condition (if non˗agentive subjects originate in a position below the phase edge), 27 Though according to Gallego and Uriagereka (2006) Phase sliding would in fact mean that TP is the phase here. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 111 nor of the Inactivity Condition (if C extracts PP out of the subject before T values the case feature on the subject). However, the situation with extraction out of postverbal subjects in Spanish is made more complex by the fact that the subject can occupy two distinct postverbal positions, as we will see in the next section. 2.2.3 A′-positions vs. A-positions: postverbal subject position in Spanish In this section we turn to a contrast that has so far not been taken into account in determining the extent to which a constituent allows extraction. With respect to the discussion of English data we have mainly been looking at preverbal subjects in spec-TP and ECM subjects in spec-VP, since these are the default positions for nominative and accusative subjects in that language, postverbal subjects being restricted to occurring in a small number of constructions (e.g. unaccusative/ passive expletive clauses). However, as already shown, Spanish subjects may be postverbal and this is not a marked pattern. Moreover, postverbal subjects may either precede or follow the object, resulting in VSO and VOS orders respectively. Rather than assuming that a postverbal subject is invariably on the edge of the v*P phase (Ordóñez 1998, Zubizarreta 1998, Ortega-Santos 2006, Jiménez-Fernández and İşsever 2012), we follow Belletti (2001, 2004) and posit that, in addition to remaining in their base position, postverbal subjects in pro-drop languages like Spanish can also occupy the specifier position of a Focus Phrase in the low vP-periphery. Specifically, we propose that the two word orders VSO and VOS arise in Spanish by two different mechanisms. VSO is the result of moving V to a higher functional head (Suñer 1992) and leaving the subject in its base position (spec-vP). By contrast, VOS comes about when V moves to a higher functional head and the subject itself is displaced to the specifier of a Focus head in the low periphery (between TP and vP); subsequently, departing slightly from Belletti (2004) and following Zubizarreta (1998) and Vicente (2009), we assume that O/IO moves to an A position. For concreteness, we take this to be the specifier of a direct/indirect Object Agreement Phrase (AgrOP/AgrIOP) below TP but above FocP, so accounting for why O precedes S.28 28 Belletti (2004) proposes remnant VP movement to this low spec-TopP in Italian. She uses binding effects as evidence in favour of her analysis: (i) a. Chi ha salutato Gianni? who have-3sg greeted Gianni ‘Who greeted Gianni?’ Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 112 Liliane Haegeman et al. Evidence that in VSO patterns the subject remains in situ comes from quantifier binding. As noted by Ordóñez (1998), a sentence-final quantified indirect object cannot bind a pronominal possessor inside a postverbal subject in the VSO structure in (62a), but when the subject occupies final position as in (62b), the bound reading is licensed: in (62a) su amigo ‘his friend’ cannot be bound by the quantifier cada ‘each’ in the direct object that follows it; by contrast, in (62b), where the indirect object precedes, this binding is licit. (62) a. *Este libro se lo regaló sui amigo [a cada niño]i. this book him it gave-3sg his friend to each boy ‘This book, his friend gave to each boy.’ (Ordóñez 1998: 318, 9c) b. Este libro se lo regaló [a cada niño]i sui amigo. this book him it gave-3sg to each boy his friend ‘This book, his friend gave to each boy.’ (Ordóñez 1998: 319, 10c) b. *Hanno [salutato Giannii ] i proprii genitori. have-3pl greeted Gianni the his.own parents ‘His own parents have greeted Gianni.’ (Belletti’s examples (46)) The object Gianni in (ib) does not c-command the subject, and thus the anaphoric subject i propri genitori ‘their own parents’ is unbound thereby violating Principle A. With respect to binding and postverbal subjects, Spanish patterns differently. In (ii) postverbal a Juan can bind the anaphoric sus propios padres (‘his parents’): (ii) Han saludado a Juani sus propiosi padres. have greeted to Juan his own parents ‘His parents have greeted Juan.’ If Belletti’s remnant VP movement were extrapolated to Spanish, we would expect sentence (ii) to induce a Principle A violation, contrary to fact. If, on the other hand, it is just the object that moves to a low spec-TopP in Spanish, sentence (ii) is predicted to be ill-formed since it occupies an A′-position and hence the subject cannot be bound. We can account for the binding effects in (ii) by assuming that the object moves to the specifier of an A-position, here termed AgrOP because this label is familiar from earlier work. Thus, the object c-commands and so can bind the subject. The idea that direct objects move to spec-AgrOP dates back to Franco (1993, 1994), who maintains that in structures where a direct or indirect object is doubled by a clitic, the clitic is an agreement marker occupying the head of the agreement projection, and the object raises to spec- AgrP. On clitic doubling, see Suñer (1988, 1992), Lyons (1990), García-Miguel (1991), Roca (1992), Torrego (1994), Sportiche (1995), Franco and Mejías-Bikandi (1997, 1999), Parodi (1998), Bleam (1999), Estigarribia (2006), Franco and Huidobro (2006), Belloro (2007), Preminger (2009), Gabriel and Rinke (2010), Ormazábal and Romero (2010), Aranovich (2011). Note incidentally that moving the object to a higher functional phrase is within the spirit of work by Ordóñez (1998) and Ortega-Santos (2006). Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 113 Assuming with Belletti that binding can be established at any stage of the derivation, the grammaticality of (62b) with the order VOS can be accounted for by supposing that the subject su amigo ‘his friend’ moves to spec-FocP, and that the Indirect Object a cada niño ‘to each boy’ moves to the specifier position of an AgrIO projection above FocP. This means that the Indirect Object ends up in an A-position from where it c-commands (and can bind) the pronoun su contained within the subject on the edge of a vP-peripheral FocP projection. Support for the claim that the indirect object occupies an A position in (62b) comes from the fact that movement across the subject su amigo does not lead to a Weak Crossover violation (see also Vicente 2009: 174–176) for recent discussion. In contrast, in (62a), with the order VSO, both subject and indirect object remain in situ within vP throughout the derivation, with the result that at no stage does the indirect object c-command (or bind) the subject. Belletti (2004) provides further evidence that postverbal subjects occupy spec-FocP in VOS structures from the observation that postverbal subjects follow adverbs (if present) that are low on the Cinque (1999) hierarchy. The same pattern holds in Spanish, as shown by examples such as (63), where the low adverb is bold-printed. (63) a. ¿De qué pintor causaron literalmente un gran of which painter caused-3pl literally a big escándalo [varios dibujos]? scandal several paintings b. *¿De qué pintor causaron un gran escándalo [varios of which painter caused-3pl a big scandal several dibujos] literalmente? paintings literally ‘Of which painter did several paintings literally cause a big scandal?’ The conclusion we draw from sentences such as (62) and (63) is that Spanish VOS subjects occupy an A-bar position in spec-FocP, whereas VSO subjects remain in situ in spec-vP. The analysis of Spanish word order outlined above has interesting implications for extraction out of subjects. Extraction out of an agentive subject in VSO structures violates the Edge Condition because agent subjects are positioned on the edge of a vP phase; however, it does not violate the Inactivity Condition if Cextracts PP from the subject before T case-marks (and thereby inactivates) the subject; nor does it violate the Freezing Principle, since the subject remains in situ. But if Spanish postverbal subjects in VOS structures move to the specifier position within a Focus Phrase position in the low periphery, such postverbal Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 120 Liliane Haegeman et al. class of violable constraints, we can differentiate between those which are relatively strong (like the Argument Condition), and those which are relatively weak (like the Freezing Condition). The strength of constraints can vary from one language to another. For example, the Preposition Stranding Constraint is an inviolable constraint in Spanish, but it is a violable constraint albeit relatively strong in English. As mentioned before, a full account of this parameterisation both in violability and in strength of a constraint requires further study. Sentences which do not violate any constraints are expected to be readily accepted by all speakers.32 But violation of a single constraint does not necessarily lead to full unacceptability, unless this is an inviolable constraint such as the Preposition Stranding Constraint in Spanish. Rather, the effect of weak constraint violations is cumulative and the ensuing degradation is proportionate to the number (and relative strength) of constraints violated. Using the traditional ?/* notation, let us suppose that (as far as violation of weak constraints is concerned) a sentence is judged ? if involving violation of a single constraint, ?? if involving a double constraint violation, ?* if involving a triple violation, and * if involving a quadruple violation. By contrast, a sentence involving violation of an inviolable constraint leads to outright unacceptability (*). Ultimately, it would be desirable of course to make this concept of graded judgments more precise and to devise an accurate notation that reflects the cumulative effect of constraint violations: see section 4.3 for discussion. In addition, it would be interesting to explore to what extent such graded judgments can be elicited experimentally (see for instance Alexopoulou and Keller 2007; Haddican 2010; also Haddican and Plunkett 2010 and references cited there; see also the discussion in Section 4.2). Sentences which violate only one weak constraint will produce a relatively low degree of unacceptability: more sensitive speakers will detect some degree ofawkwardness, whereas more tolerant speakers may simply accept them (so accounting for inter-speaker variation in judgments), with the result that the perceived level of degradation may vary from one speaker to another. We account for this by supposing that speakers have differing tolerance thresholds (in the sense that some speakers may be more tolerant of particular constraint violations than others). 32 Note that this does not mean that they should be expected to receive a perfect score (e.g. 7 on a 7-point Likert scale), since experimental research suggests that any extraction of one constituent out of another leads to some lowering of acceptability, because of the problems faced by the parser in locating the gap associated with the filler (i.e. the moved constituent): furthermore the greater the distance between filler and gap, the lower the acceptability level is likely to be. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 121 In the next two sections, we will show how our analysis accounts for the gradient acceptability of extraction of various types of constituent out of various types of subject in Spanish (Section 3.2) and English (Section 3.3). An important point to note at the outset, however, is that our judgments in Sections 3.2 and 3.3 below may differ from the judgments of other linguists reported in earlier sections. This is because differences tend to be represented in black and white terms when only two sentences (e.g. a minimal pair) are compared, but often turn out toinvolve shades of grey when a larger set of sentences are compared. By way of illustration, consider extraction from subjects with and without preposition piedpiping in English. The pied-piping examples are conventionally treated as grammatical, and their stranding counterparts as ungrammatical. However, Jurka (2010) reported that although stranding examples like (43b) received a low acceptability rating of only 2.51 on a 7-point scale, pied-piping examples like (43a) received a score which was very little higher (3.29), suggesting that pied-piping examples are far from fully acceptable. 3.2 Extraction from subjects in Spanish In the light of the assumptions made above, let’s take another look at the varying degrees of degradation resulting from extraction out of subjects in pre- andpost-verbal positions in Spanish. An important point to be borne in mind isthat we assume here that – in consequence of the Extraction Constraint (31) – extraction is launched from the superficial position of the constituent. We continue to adhere to this position throughout our discussion of English and Spanish in the remainder of the paper. 3.2.1 Structures which violate no constraint Consider sentence (56b): (56) b. ¿De qué príncipe fueron publicadas [varias fotos of which prince were-3pl published several photos comprometedoras]? compromising ‘Of which prince were several compromising photos published?’ In this example, extraction is launched from within a postverbal passive theme subject, which might either be positioned in situ in comp-VP, or have raised to spec-FocP (see Section 2.1.4). In neither case will there be a violation of the Edge Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 122 Liliane Haegeman et al. Condition (because the subject is not positioned on the edge of a phase), or of the Preposition Stranding Constraint (because the preposition has not been stranded), or the Specificity Condition (because the subject is non-specific), or of the Argument Condition (because the extractee is an argument). Nor will there be any violation of the Inactivity Condition: this is because, if the subject moves to spec-FocP, it will not be in an A-position, and if it remains in situ in comp-VP it remains active (and hence transparent for extraction) if C probes to extract the PP before T probes to value the case feature on the subject. If the subject moves to spec-FocP, extracting PP from it will result in violation of the Freezing Principle, but this only leads to a mild degradation; however, if the subject remains in situ in comp-VP, there will be no freezing violation (nor any other constraint violation). Accordingly, (56b) would be expected to be judged to be fine, since it has one derivation on which the subject remains in comp-VP, and no constraint is violated. 3.2.2 Structures violating a single constraint Structures which violate an inviolable constraint (such as the P-stranding constraint in Spanish) are severely degraded, as we see from our earlier example (56a): (56) a. *¿Qué príncipe fueron publicadas [varias fotos which prince were-3pl published several photos comprometedoras de]? compromising of? ‘Which prince were several compromising photos of published?’ By contrast, the violation of a weak constraint will lead to weak degradation, as can be illustrated in relation to (60a): (60) a. ?¿De qué coches parece que fueron arrestados of which cars seem-3sg that were-3pl arrested [los conductores]? the drivers The derivation of (60a) is essentially parallel to that of (56b), save for the fact that extraction in (56b) takes place out of a non-specific subject, whereas that in (60a) takes place out of a specific subject. Consequently, since (60a) involves violation of a single weak constraint, it leads to a mild degradation. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 123 A similar weak constraint violation also arises in cases such as (76): (76) a. ?¿De qué coche han ganado [varios pilotos] dos carreras? of which car have-3pl won several drivers two races ‘Of which car have several drivers won two races?’ b. ?¿De qué equipo han protestado [muchos jugadores]? of which team have-3pl protested many players ‘Which team have many players in protested?’ (Gallego 2007: 293, fn.8) Following our discussion in Section 2.1.4, we assume that the bracketed subject inexamples like (76) remains in situ, and that the agent subject of a transitive orunergative predicate is merged in a position on the edge of a vP phase (e.g. spec-VoiceP). This in turn means that extraction out of an in situ agentive subject will give rise to violation of the Edge Condition. However, this is a weak constraint and because no other constraint is violated in (76), the resulting sentences show only mild degradation. 3.2.3 Structures violating multiple constraints Now consider (77), which Uriagereka (1988: 122) judges to be doubly degraded (??): (77) ??¿De qué artistas han herido tu sensibilidad [las of which artists have-3pl hurt your sensitivity the obras]? works ‘Which artists have the works of hurt your sensitivity?’ In this type of VOS structure, the subject raises to the edge of FocP, and the objectraises to the edge of a superordinate AgrOP projection. Since the subject is specific and has moved to spec-FocP, extraction from it violates both the FreezingPrinciple and the Specificity Condition, causing the sentence to be double degraded. A similar double degradation is found in (78): (78) ??¿De qué coche han provocado [los pilotos] un accidente? of what car have-3pl caused the drivers an accident ‘Of which car have the drivers caused an accident?’ Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 124 Liliane Haegeman et al. (78) displays VSO order, so we assume that the subject remains in its initial merge position, spec-vP. Here, extraction takes place from a bracketed in situ specific agentive subject on the edge of a phase. This results in a violation of both the Edge Condition and the Specificity Condition, and the resulting double constraint violation leads to double degradation. More degraded still is (79): (79) ?*¿De qué electrodoméstico parece que [el inventor] of which electrical.appliance seem-3sg that the inventor causó tanta conmoción? caused-3sg such commotion ‘Which electrical appliance does it seem that the inventor of caused a big stir?’ Here, the italicized argumental PP is extracted out of a specific subject in spec-TP. Since the subject has moved to spec-TP, extraction incurs violation of the Freezing Principle. Since the subject is inactive once it reaches spec-TP (by virtue of having had its case feature valued), extraction also leads to violation of the Inactivity Condition. In addition, since the subject is specific, extraction also violates the Specificity Condition. This triple constraint violation leads to triple degradation (?*). Observe that in (79) the extractee is D-linked, which mitigates the severity of the violation. As seen in (80), the degradation becomes even more severe if the extracted PP is not D-linked: (80) *¿De qué parece que [el inventor] causó tanta of which seem-3sg that the inventor caused-3sg such conmoción? commotion ‘What does it seem that the inventor of caused a big stir?’ A parallel quadruple constraint violation is also found in (81): (81) *¿Con qué tipo de cicatriz en la cara dijiste que [el hombre] with what type of scar on the face said-2sg that the man asesinó a una señora anciana? murdered-3sg to a lady old ‘With what kind of scar on his face did you say that [the man] murdered an old lady?’ Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 125 This is because, by virtue of extracting an adjunct out of a specific, moved, inactive subject at the head of an A-chain, (81) involves a quadruple (Freezing, Inactivity, Specificity and Argument) violation, so leading to extremely severe degradation (especially as the Argument Constraint is a strong one). 3.3 Extraction from subjects in English In this section, we return to extraction out of subjects in English. Our discussion will be brief, partly to avoid unnecessary repetition, and partly because (as noted earlier), the possibilities for extracting out of subjects are more limited in English than in Spanish, because English subjects canonically raise to spec-TP. However, as we saw in Section 1.1.1, extraction is possible from the base position of a subject in existential expletive structures such as: (4) a. Which candidate were there [posters of ] all over the town? (Lasnik and Park 2003: 651) This is because, under our analysis, no violation of any constraint occurs in such structures: e.g. there is no Freezing violation because the subject remains in situ within VP, no Edge violation because the subject is not positioned on the edge of a phase, no Inactivity violation if C extracts PP before T values the case feature on the subject, and so on. As illustrated in Section 3.3 for Spanish, each additional constraint violation leads to further degradation. In this connection, consider the contrast below: (82) a. ?? Of which famous celebrity were [compromising pictures] published? b. ?* Which famous celebrity were [compromising pictures of ] published? In both cases, the subject has moved to spec-TP and had its case feature valued as nominative. Accordingly, both sentences involve Freezing and Inactivity violations. However, (82a) is even more degraded since it also violates the Preposition Stranding Constraint: thus (82a) is doubly degraded and (82b) triply degraded – in line with the experimental results for sentences like (43) reported by Jurka (2010). We find even more severe degradation in sentences such as (83): (83) * Who were [compromising pictures of ] published? Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 126 Liliane Haegeman et al. As in (82b), extraction out of the subject in spec-TP leads to Freezing, Inactivity and Preposition Stranding violations; however, the fact that the subject is not D-linked means that (83) is even more degraded than (82). Observe that so far, in keeping with the literature, we have treated the D-linking Condition in (75) as an “amelioration” condition. An alternative possibility would be to treat it as a “degradation” condition such as (84): (84) D-Linking Condition (reformulated as a degradation condition) Extraction of a (pro)nominal which is not D-linked is degraded. It would then be the case that (84) involves a quadruple (Freezing, Inactivity, Preposition Stranding and D-linking) violation, so accounting for its severe degradation. The reformulation in (84) will, however, have further ramifications that we cannot pursue here. Having seen how our analysis accounts for differing degrees of degradation in different types of structure, in the next section we turn to address a number of issues which arise from it. 4 Issues arising from our analysis Our account raises three important issues.33 One concerns the nature of the constraints we posit. The second is whether these constraints are ultimately reducible to non-syntactic factors (e.g. processing considerations). The third is whether our assumption that different types of constraint have different degrees of strength is equivalent to importing into the system the concept of constraintranking familiar from Optimality Theory/OT. We address each of these questions in the three subsections below. 4.1 The nature of the constraints The key point of our paper is that the Subject Condition is not a primitive constraint in itself, but rather is reducible to the cumulative interaction of a number of more general constraints which are not specific to subjects. However, this raises the questions of what the nature of these more general constraints is, why they should exist, and why they should have the properties that they do. 33 We thank the anonymous reviewers for bringing up these points. Their comments have enabled us to sharpen the discussion of the issues. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 127 Traditional work conceived of island constraints as derivational conditions which prevent movement operations extracting constituents out of opaque domains in the syntax. However, this derivational view has been challenged by evidence that (certain types of) constraint violation can be repaired by ellipsis operations (Merchant 1999, 2001, 2002, 2004, 2006, 2008; Lasnik 2001a; Fox and Lasnik 2003; Boeckx and Lasnik 2006; Bošković 2011) or by resumption (Boeckx 2001, 2003, 2008, 2012; Aoun et al. 2001; McCloskey 2002). This has led to the suggestion that repairable constraints are PF interface conditions barring traces from occurring in opaque domains at PF, and that they can be circumvented either if the trace is spelled out as a resumptive34, or if the trace is obliterated by ellipsis of some constituent containing the trace.35 This raises the question of whether the constraints central to our analysis are derivational or interface conditions. To elucidate the issue, let’s consider the question of whether the Preposition Stranding Constraint/PSC in English is a derivational constraint or a PF Interface Condition/PFIC. If PSC is a PF-interface locality condition and if locality constraints are representational in nature (Bošković 2011: 6), PSC can be formulated in representational terms as a PFIC barring PF representations in which a PP contains a trace which is bound from outside a superordinate XP which itself binds a trace. Such a PFIC account of PSC will correctly specify that an unsluiced structure like (85a) induces a PSC violation, but its sluiced counterpart (85b) does not: (85) a. * At the rally, supporters of one of the speakers were arrested, but I’m not sure who supporters of were arrested. b. At the rally, supporters of one of the speakers were arrested, but I’m not sure who. The underlined clause in (85a) has the structure (86): 34 But see Polinsky et al. (in press) on resumption in English. 35 Richards (2011) argues that the repair property of ellipsis can be accounted for by positing that elided constituents are transferred only to the LF interface, not to the PF interface (hence they will not violate PF constraints). Boeckx (2008, 2012) argues that it is not Sluicing per se which repairs constraint violations, but rather the presence of a null resumptive pronoun ( pro), e.g. in a structure like: (i) John countered the claim that Fido bit someone, but I can’t remember who John countered the claim that Fido bit pro (cf. Boeckx 2012: 93, 22) Jiménez-Fernández (2009) argues that island constraints are both syntactic and interfaceconnected. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 128 Liliane Haegeman et al. (86) *[CP whoi [C ø] [TP [QPj supporters [PP of ti ]] [T were] arrested tj ]] The PSC violation arises in (86) because the of-PP contains a trace of who which is bound from outside the containing QP supporters of ti, and this containing QP binds a trace of itself in comp˗VP. As is expected if PSC is a PF interface condition, violation of PSC is obviated by Sluicing of the TP supporters of t were arrested, because Sluicing erases the traces which give rise to the PSC violation. Interestingly, an alternative way of obviating PSC violations found in colloquial English is the use of a resumptive pronoun in structures like (87): (87) Ancelotti played the three, which we thought [one of them] might be sacrificed. (Gabby Logan, BBC Radio 5) This is in keeping with a substantial body of research suggesting that resumption can be used to repair PF locality constraints (Pesetsky 1997, 1998; Aoun and Li 2003; Boeckx 2012). Having looked at the nature of PSC, let’s now turn to consider more briefly the nature of some of the other constraints central to our analysis. The Edge Condition and the Inactivity Condition are formulated by Chomsky (2008) as derivational conditions imposing locality restrictions on the search domain for a probe (i.e. to limit how far a probe can “look” into a syntactic structure it its search for a goal). However, sentences like (88) below suggest that Sluicing can repair violation of these conditions: (88) a. The biography of a famous A-list Hollywood celebrity caused a scandal. Guess who! b. * Guess who the biography of caused a scandal? In this example interrogative who is interpreted as the complement of a subject internal PP, corresponding to of a famous A-list Hollywood celebrity. Extraction of who from the subject in (88b) leads to a quadruple violation: (87b) violates the Freezing Condition (69), the Inactivity Condition (72), the Preposition Stranding Condition (71) and the Specificity Condition (73). And yet the ellipsis of TP including the offending subject leads to a grammatical result in (87a). This would suggest that such locality constraints are repairable PF interface conditions rather than irreparable derivational conditions. Now consider the Specificity Condition (73), the Argument Condition (74), and the D-linking Condition (75). We argued in sections 1.3.2 and 1.3.3 that these are ultimately reducible to the Intervention Condition (48). Since intervention ef- Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 129 fects are sensitive to the semantic properties of the mover and the intervener, it might at first sight seem plausible (on conceptual grounds) to suppose that the Intervention Constraint is an interface condition on LF representations, barring an intervention-sensitive constituent from binding a trace across a relevant type of intervener at LF. If so, we would expect that intervention effects cannot be alleviated by Sluicing, since Sluicing is a PF operation, and the output of PF operations is invisible at LF. However, any such claim would be challenged by the observation made in Chung et al. (1995) that Sluicing can repair wh-island violations, e.g. in sentences such as: (89) Sandy was trying to work out which students would be able to solve a certain problem, but she wouldn’t tell us which one (Chung et al. 1995, 84a) The repairability of wh-intervention violations suggests that the Intervention Constraint is a PF Interface Condition/PFIC. The general conclusion which our brief discussion in this section leads us tois that the constraints central to our analysis are probably PFICs. However, thisraises the question of the nature of the relevant PFICs and to what extent PFICs are relevant in the syntax. Bošković (2011), elaborating on earlier work byChomsky (1972), proposes that locality violations incurred in a derivation result in the marking (in the syntax) of the element that is responsible for blockingmovement. He does this concretely by using the diacritic *. With respect to illicit extraction from an island, he proposes that when a wh-moved element crosses an island boundary, the island is *-marked. In other words, the diacritic * is assigned to the element that has caused a locality-of-movement violation. Thepresence of a diacritic * in the final PF representation of a derivation leads toa crash. How ever, such a violation is “repaired” (i.e., does not occur) if the *-marked element is deleted at PF since in that case, no * is present in the final PF representation. In the next section, we consider whether the relevant PF interface conditions are reducible to processing constraints designed to increase the parsability of PF structures. 4.2 Processing factors There is evidence from a range of studies (including Kluender 1998, 2004, 2005; Kluender and Kutas 1993; Hofmeister 2007, 2008, 2011; Sag et al. 2007; Hofmeister et al. 2007; Hofmeister and Sag 2010; Clausen 2010; Chaves 2013) that processing Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 136 Liliane Haegeman et al. (68) Extraction Constraint (= 31) Extraction is only possible from the head of a chain. (69) Freezing Principle (= 3) A moved constituent is frozen for extraction. (70) Edge Condition (= 8) The edge of a phase is opaque for extraction. (71) Preposition Stranding Condition/PSC (= 54) No preposition can be stranded (% inside a moved constituent). (72) Inactivity Condition (= 11) An inactive A-chain (i.e. one with no unvalued A-features) is opaque for extraction. (73) Specificity Condition (= 33) Specific nominals are opaque domains for extraction. (74) Argument Condition (= 46) Extraction is degraded when the extractee is not an argument. In addition, we have shown how the D-linking effect can either be handled in terms of an amelioration condition like (75), or as a degradation condition like (84): (75) D-linking Condition (formulated as an “amelioration” condition) Extraction is ameliorated when the extractee is D-linked. (84) D-Linking Condition (formulated as a “degradation” condition) Extraction of a (pro)nominal which is not D-linked is degraded. Decomposing the Subject Condition into a set of subconditions whose effect is cumulative allows a more fine-grained scale of judgments: the more conditions which are violated the more degraded the outcome is (in conformity with the assumptions made in Linear Optimality Theory). To account for inter-speaker variation reported in the literature we postulated that different speakers have different tolerance thresholds. It is to be hoped (and expected) that future theoretical developments will lead to further decomposition of constraints like those in (68– 75), so that even the principles we have listed here end up being decomposed intoprimitive constraints. Any such development will be beneficial and will make a more refined account of grammaticality judgements possible. Needless to say, this would not change the general line of our argumentation. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 137 Acknowledgments: Authors’ names are listed in alphabetical order. We are grateful to Bob Borsley, Will Harwood, Philip Hofmeister and Louisa Sadler for helpful comments on an earlier version of this paper. Haegeman’s research is funded by FWO: 2009-Odysseus-Haegeman-G091409. Examples cited from other works are adapted in inconsequential typographical ways to fit in with the style of presentation (and the example numbering) adopted here. Grammaticality judgments for sentences cited from other works are those given by the linguists cited. In response to comments from anonymous reviewers some preliminary remarks are in order to clarify our position. In this paper our approach is essentially syntactic in nature, although at the end of the paper we briefly address alternative (e.g. pragmatic and processing) perspectives. Our data are based on introspective acceptability judgments: for a defense of the use of introspective data and a comparison with usage-based data see Newmeyer (2003, 2005, 2006a, 2006b), and for a comparison between introspective and experimental data see Sprouse (2011), Sprouse et al. (2011), Sprouse and Almeida (2011, 2012a, 2012b), and Schütze and Sprouse (2012). We adopt an approach to syntax which presupposes movement operations, but is relatively theory-neutral in respect of notbeing tied to a specific (e.g. Minimalist or Cartographic) implementation. Derivations and structural representations are simplified by showing only essential details (and sometimes omitting null constituents); trace copies of moved constituents are shown as t, in order to save space. References Abels, Klaus. 2003. Successive cyclicity, antilocality and adposition stranding. Storrs, CT:University of Connecticut dissertation. Aboh, Enoch. 2004. The morphosyntax of complement-head sequences. Clause structure and word order patterns in Kwa. Oxford: Oxford University Press. Abney, Steven P. 1987. The English noun phrase in its sentential aspect. Cambridge, MA: MITdissertation. Ackema, Peter. 2008. Restrictions on subject extraction: A PF interface account. Ms., University of Edinburgh. Adli, Aria. 2011. A heuristic mathematical approach for modelling constraint cumulativity: Contrastive focus in Spanish and Catalan. The Linguistic Review 28. 111–173. Agbayani, Brian. 2000. Wh-Subjects in English and the vacuous movement hypothesis. Linguistic Inquiry 31. 703–730. Aoun, Joseph and Yen-Hui A. Li. 2003. Essays on the representational and derivational nature ofgrammar: The diversity of wh-constructions. Cambridge MA: MIT Press. Aoun, Joseph, Lina Choueiri and Norbert Hornstein. 2001. Resumption, movement, and derivational economy. Linguistic Inquiry 32. 371–403. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 138 Liliane Haegeman et al. Aguero-Bautista, Calixto. 2012. Non-native acquisition and language design. In Anna Maria Di Sciullo (ed.), Towards a biolinguistic understanding of Grammar. Essays on interfaces, 217–238. Amsterdam: John Benjamins. Aissen, Judith. 1998. Markedness and subject choice in Optimality Theory. Ms., UC Santa Cruz. Alboiu, Gabriela. 2002. The features of movement in Romanian. Bucharest: Editura Universitätii Bucureşti. Alexiadou, Artemis, Liliane Haegeman and Melita Stavrou. 2007. Noun phrase in the generative perspective. Berlin: Mouton de Gruyter. Alexopoulou, Theodora and Frank Keller. 2007. Locality, cyclicity and resumption: At the interface between the grammar and the human sentence processor. Language 83. 110–160. Alrenga, Peter. 2005. A sentential subject asymmetry in English and its implications for complement selection. Syntax 8(3). 175–207. Aranovich, Roberto. 2011. Optional agreement and grammatical functions: A corpus study of dative clitic doubling in Spanish. Pittsburgh, PA: University of Pittsburgh dissertation. Bagchi, Tista. 2007. On theta assignment assignment by feature checking. In Eric Reuland, Tanmoy Bhattacharya and Giorgos Spathas (eds.), Argument structure. 159–175. Amsterdam: John Benjamins. Baunaz, Lena. 2011. The grammar of French quantification (Studies in natural language and linguistic theory 83). Dordrecht: Springer. Baunaz, Lena. 2012. The French complementiser que under scrutiny. Ms., University of Geneva. Belletti, Adriana. 2001. Inversion as focalisation. In Aafke Hulk and Jean-Yves Pollock (eds.), Subject inversion in Romance and the theory of Universal Grammar. 60–90. Oxford: Oxford University Press. Belletti, Adriana. 2004. Aspects of the low IP area. In Luigi Rizzi (ed.), The structure of CP and IP. 16–51. Oxford: Oxford University Press. Belloro, Valeria A. 2007. Spanish clitic doubling: A study of the syntax-pragmatics interface. Buffalo, NY: State University of New York dissertation. Bianchi, Valentina and Cristiano Chesi. 2006. Phases, left branch islands, and computational nesting. UPenn Working Papers in Linguistics 12: Proceedings of the 29th Penn Linguistic Colloquium. 15–28. Bianchi, Valentina and Cristiano Chesi. 2012. Subject islands and the subject criterion. In Valentina Bianchi and Cristiano Chesi (eds.), Enjoy Linguistics! Papers offered to Luigi Rizzi on the occasion of his 60th birthday. 25–53. Siena: CISCL Press. http://www.ciscl.unisi.it/ gg60/ (accessed 13 February 2013). Bleam, Tonia M. 1999. Leísta Spanish and the syntax of clitic doubling. Newark, DA: University of Delaware dissertation. Boeckx, Cedric. 2001. Mechanisms of chain formation. Storrs, CT: University of Connecticut dissertation. Boeckx, Cedric. 2003. Islands and chains. Stranding as resumption. Amsterdam: John Benjamins. Boeckx, Cedric. 2008. Bare syntax. Oxford: Oxford University Press. Boeckx, Cedric. 2012. Syntactic Islands. Cambridge: Cambridge University Press. Boeckx, Cedric and Howard Lasnik. 2006. Intervention and repair. Linguistic Inquiry 37. 150–155. Bošković, Željko. 2011. Rescue by PF deletion, traces as (non)interveners, and the that-trace effect. Linguistic Inquiry 42. 1–44. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 139 Bresnan, Joan. 1994. Locative inversion and the architecture of Universal Grammar. Language 70. 72–131. Bresnan, Joan. 2000. Optimal syntax. In Joost Dekkers, Frank Van der Leeuw and Jeroen Van de Weijer (eds.), Optimality theory: Phonology, syntax and acquisition, 334–385. Oxford: Oxford University Press. Bresnan, Joan. 2001. Explaining morphosyntactic competition. In Mark Baltin and Chris Collins (eds.), The Handbook of contemporary syntactic theory. 11–44. Oxford: Blackwell. Broekhuis, Hans. 2006. Extraction from subjects: Some remarks on Chomsky’s “On Phases”. InHans Broekhuis, Norbert Corver and Riny Hybreghts (ed.), Organising Grammar, 59–68. Berlin: Mouton de Gruyter. Broekhuis, Hans. 2008. Derivations and Evaluations: Object shift in the Germanic languages. Mouton: Berlin. Campbell, Richard. 1996. Specificity operators in SpecDP. Studia Linguistica 50. 161–188. Cattell, Ray. 1976. Constraints on movement rules. Language 52. 18–50. Chaves, Rui. 2012. On the grammar of extraction and coordination. Natural Language and Linguistic Theory 30. 465–512. Chaves, Rui. 2013. An expectation-based account of subject islands and parasitism. Journal ofLinguistics 49(2). 285–327. Choi, Hye-Won. 1996. Optimizing structure in context: Scrambling and information structure. Stanford, CA: Stanford University dissertation. Chomsky, Noam. 1964. The logical basis of linguistic theory. In Horace G. Lunt (ed.), Proceedings of the Ninth International Congress of Linguists, 914–978. The Hague: Mouton and Co. Chomsky, Noam. 1972. Some empirical issues in the theory of transformational grammar. In Stanley Peters (ed.), Goals of linguistic theory, 63–130. Englewood Cliffs, NJ: Prentice Hall. Chomsky, Noam. 1973. Conditions on transformations. In Stephen R. Anderson and Paul Kiparsky (eds.), A festschrift for Morris Halle. 232–286. New York: Holt, Rinehart and Winston. Chomsky, Noam. 1977. On Wh-movement. In Peter W. Culicover, Thomas Wasow and Adrian Akmajian (eds.), Formal syntax. 71–132. New York: Academic Press. Chomsky, Noam. 1981. Lectures on government and binding. Foris: Dordrecht. Chomsky, Noam. 1986. Barriers. Cambridge, MA: MIT Press. Chomsky, Noam. 2000. Minimalist inquiries. In Roger Martin, David Michaels and Juan Uriagereka (eds.), Step by step: Essays on minimalist syntax in honor of Howard Lasnik. 89–155. Cambridge, MA: MIT Press. Chomsky, Noam. 2001. Derivation by phase. In Michael Kenstowicz (ed.), Ken Hale: A life in language, 1–52. Cambridge, MA: MIT Press. Chomsky, Noam. 2005. Three factors in language design. Linguistic Inquiry 36. 1–22. Chomsky, Noam. 2007. Approaching UG from below. In Uli Sauerland and Hans-Martin Gärtner (eds.), Interfaces + Recursion = Language? Chomsky’s minimalism and the view from syntax-semantics. 1–30. Berlin: Mouton de Gruyter. Chomsky, Noam. 2008. On phases. In Robert Freidin, Carlos P. Otero and María-Luisa Zubizarreta (eds.), Foundational issues in linguistic theory. 133–166. Cambridge, MA: MITPress. Chomsky, Noam. 2013. Problems of projection. To appear in Lingua. Chomsky, Noam and Howard Lasnik. 1993. The theory of principles and parameters. In Joachim Jacobs, Arnim von Stechow, Wolfgang Sternefeld and Theo Venneman (ed.), Syntax: Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 140 Liliane Haegeman et al. Aninternational handbook of contemporary research, Vol. 1, 506–569. Berlin: Mouton deGruyter. Chung, Sandra. 1994. Wh-agreement and “referentiality” in Chamorro. Linguistic Inquiry 25. 1–44. Chung, Sandra and James McCloskey. 1983. On the interpretation of certain island facts in GPSG. Linguistic Inquiry 14. 703–714. Chung, Sandra, William Ladusaw and James McCloskey. 2011. Sluicing: Between structure and inference. In Rodrigo Gutiérrez-Bravo, Line Mikkelsen and Eric Potsdam (eds.), Representing language: Essays in honor of Judith Aissen, 31–50. University of California, Santa Cruz: Linguistic Research Center. Cinque, Guglielmo. 1977. Towards a unified treatment of island constraints. In Wolfgang Dressler and Wolfgang Meid (ed.), Proceedings of the twelfth international congress of linguists. 344–348. Innsbruck: Innbrucker Beitrage zur Sprachwissenschaft. Cinque, Guglielmo. 1990. Types of Aʹ-dependencies. Cambridge, MA: MIT Press. Cinque, Guglielmo. 1999. Adverbs and functional heads. Oxford: Oxford University Press. Cinque, Guglielmo. 2000. On Greenberg’s Universal 20 and the Semitic DP. University of Venice Working Papers in Linguistics 10. 45–61. Clausen, David R. 2010. Processing factors influencing acceptability in extractions from complex subjects. Ms., Stanford University. Collins, Chris. 2007. Home sweet home. NYU Working Papers in Linguistics 1. 1–34. Collins, Chris and Andrew Radford. 2012. Gaps, ghosts and gapless relative clauses in spoken English. Ms., New York University and University of Essex. To appear in Studia Linguistica. Comorovski, Ileana. 1996. Interrogative phrases and the syntax-semantics interface. Cambridge, MA: MIT Press. Contreras, Heles. 1987. Small clauses in Spanish and English. Natural Language and Linguistic Theory 5. 225–244. Cresti, Diana. 1995. Extraction and reconstruction. Natural Language Semantics 3. 79–122. Cruschina, Silvio. 2012. Discourse-related features and functional projections. Oxford: Oxford University Press. Davies, William D. and Stanley Dubinsky. 2003. On extraction from NPs. Natural Language and Linguistic Theory 21. 1–37. Deal, Amy R. 2009. The origin and content of expletives: Evidence from selection. Syntax 12. 285–323. Deane, Paul D. 1991. Limits to attention: A cognitive theory of island phenomena. Cognitive Linguistics 2. 1–63. Dekkers, Joost. 1999. Derivations and evaluations: On the syntax of subjects and complementizers. Amsterdam: University of Amsterdam/HIL dissertation. Dekkers, Joost, Frank van der Leeuw and Jeroen van der Weijer (eds.). 2000. Optimality theory: Phonology, syntax and acquisition. Oxford: Oxford University Press. Diesing, Molly. 1990. The syntactic roots of semantic partition. Amherst, MA: University of Massachusetts dissertation. Diesing, Molly. 1992. Indefinites. Cambridge MA: MIT Press. Drummond, Alex, Norbert Hornstein and Howard Lasnik. 2010. A puzzle about P-stranding and a possible solution. Linguistic Inquiry 41(4). 689–692. Emonds, Joseph E. 1976. A transformational approach to English syntax. New York: Academic Press. Enç, Mürvet. 1991. The semantics of specificity. Linguistic Inquiry 22. 1–25. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 141 Endo, Yoshio. 2007. Locality and information structure. A cartographic approach to Japanese. Amsterdam: John Benjamins. Engdahl, Elisabet. 1980. The syntax and semantics of questions in Swedish. Amherst, MA:University of Massachussetts dissertation. Erteschik-Shir, Nomi. 1973. On the nature of island constraints. Cambridge, MA: MIT dissertation. Erteschik-Shir, Nomi. 1981. On extraction from noun phrases (picture noun phrases). In Adriana Belletti, Luciana Brandi and Luigi Rizzi (eds.). Theory of markedness in Generative Grammar; Proceedings of the 4th GLOW Conference, 147–169. Pisa: Scuola Normale Superiore. Erteschik-Shir, Nomi. 2006. What’s what? In Gisbert Fanselow, Caroline Féry, Ralf Vogel and Matthias Schlesewsky (eds.), Gradience in grammar. 317–335. Oxford: Oxford University Press. Erteschik-Shir, Nomi. 2007. Information structure: The syntax-discourse interface. Oxford: Oxford University Press. Erteschik-Shir, Nomi and Shalom Lappin. 1979. Dominance and the functional explanation ofisland phenomena. Theoretical Linguistics 6. 41–86. Estigarribia, Bruno. Why clitic doubling? A functional analysis for Rioplatense Spanish. In Timothy L. Face and Carol A. Klee (eds.), Selected Proceedings of the 8th Hispanic Linguistics Symposium, 123–136. Somerville, MA: Cascadilla Proceedings Project. Fábregas, Antonio and Ángel L. Jiménez-Fernández. 2012. Extraction from fake adjuncts and first phase syntax. Poster presented at Going Romance, KU Leuven, 1–8 December. Fassi Fehri, Abdulkader. 1999. Arabic modifying adjectives and DP structures. Studia Linguistica 53. 105–154. Fiengo, Robert and James Higginbotham. 1981. Opacity in NP. Linguistic Analysis 7. 347–373. Fodor, Jerry. 1983. Phrase structure parsing and the island constraints. Linguistics and Philosophy 6. 163–223. Fox, Danny and Howard Lasnik. 2003. Successive-cyclic movement and island repair: The difference between sluicing and VP ellipsis. Linguistic Inquiry 34. 143–154. Frampton, John. 1991. Relativized minimality: A review. The Linguistic Review 8. 1–46. Franco, Jon. 1993. On object agreement in Spanish. Los Angeles, CA: University of Southern California dissertation. Franco, Jon. 1994. The morphological, syntactic and semantic agreement status of Spanish object clitics. ASJU XXVIII. 707–789. Franco, Jon and Errapel Mejías-Bikandi. 1997. Overt and Covert Raising to Spec AGR. Linguistic Analysis 27. 79–107. Franco, Jon and Errapel Mejías-Bikandi. 1999. The Presuppositionality Condition and Spanish clitic-doubled objects. In Jean-Marc Authier, Barbara Bullock and Lisa Reed (eds.), Formal Perspectives on Romance Linguistics. 107–119. Amsterdam/Philadelphia: John Benjamins. Franco, Jon and Susana Huidobro. 2006. Ethical datives, clitic doubling and the theory of pro. Selected Proceedings of the 10th Hispanic Linguistics Symposium, ed. Joyce Bruhn de Garavito and Elena Valenzuela, 215–224. Somerville, MA: Cascadilla Proceedings Project. Frascarelli, Mara and Ángel L. Jiménez-Fernández. 2012. Specificity effects at the discoursegrammar interface. Paper presented at Going Romance, University of Leuven, 1–8 December. Frascarelli, Mara and Ángel L. Jiménez-Fernández. 2013. Contrast at the interfaces. Paper presented at Cambridge Comparative Syntax 2, University of Cambridge, 10–11 May. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 142 Liliane Haegeman et al. Frazier, Lyn and Charles Clifton, Jr. 2002. Processing “d-linked” phrases. Journal of Psycholinguistic Research 31. 633–660. Friedin, Robert. 1978. Cyclicity and the theory of grammar. Linguistic Inquiry 9. 519–549. Friedmann, Naama, Adriana Belletti and Luigi Rizzi. 2009. Relativized minimality: Types of intervention in the acquisition of A-bar dependencies. Lingua 119. 67–88. Gabriel, Christoph and Esther Rinke. 2010. Information packaging and the rise of clitic doubling in the history of Spanish. In Gisella Ferraresi and Rosemarie Lühr (eds.), Diachronic studies on information structure: Language acquisition and change. 55–77. Berlin: Mouton de Gruyter. Gallego, Ángel. 2007. Phase Theory and Parametric Variation. Barcelona: Universitat Autònoma de Barcelona dissertation. Gallego, Ángel. 2009. On freezing effects. Iberia:IJTL 1. 33–51. http://www.siff.us.es/iberia/ index.php/ij/article/view/11/14 (accessed 17 October 2013). Gallego, Ángel. 2011. Successive cyclicity, phases, and CED effects. Studia Linguistica 65. 32–69. Gallego, Ángel and Juan Uriagereka. 2006. Sub-extraction from subjects: A phase-based theory. Ms., Universitat Autònoma de Barcelona and University of Maryland. Gallego, Ángel and Juan Uriagereka. 2007. Conditions on sub-extraction. In Luis Eguren and Olga Fernández-Soriano (eds.), Coreference, modality, and focus. 45–70. Amsterdam/ Philadephia: John Benjamins. García-Miguel, José María. 1991. La duplicación del objeto directo e indirecto como concordancia. Verba 18. 375–410. Gibson, Edward. 1998. Linguistic complexity: Locality of syntactic dependencies. Cognition 68. 1–76. Goldberg, Adele E. 2006. Constructions at work: The nature of generalization in language. Oxford: Oxford University Press. Goodall, Grant. 2004. On the syntax and processing of wh-questions in Spanish. WCCFL 23 Proceedings. In Benjamin Schmeiser, Vineeta Chand, Ann Kelleher and Angelo Rodriguez (eds.). 101–114. Somerville, MA: Cascadilla Press. Grimshaw, Jane. 1990. Argument structure. Cambridge, MA: MIT press. Grimshaw, Jane. 1991. Extended Projection. Ms., Brandeis University. Grimshaw, Jane. 1997. Projection, heads and optimality. Linguistic Inquiry 28. 373–422. Grimshaw, Jane. 2000. Locality and Extended Projection. In Peter Coopmans, Martin Everaert and Jane Grimshaw (eds.), Lexical specification and insertion. 115–134. Amsterdam: John Benjamins. Grimshaw, Jane. 2005. Words and structure. Stanford, CA: CLSI Publications. Grohmann, Kleanthes K. 2003. Prolific domains: On the anti-locality of movement dependencies. Amsterdam: John Benjamins. Grohmann, Kleanthes K. 2010. Anti-locality: Too-close relations in grammar. Ms., University ofCyprus. Haddican, William. 2010. Theme-goal ditransitives and theme passivisation in British English dialects. Lingua 120. 2424–2443. Haddican, William and Bernadette Plunket. 2010. Formalising syntactic variability: Context and issues. Lingua 120(5). 1057–1061. Haegeman, Liliane. 2012. Adverbial clauses, main clause phenomena and composition of the left periphery. Oxford: Oxford University Press. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 143 Haegeman, Liliane and Barbara Ürögdi. 2010a. Referential CPs and DPs: An operator movement account. Theoretical Linguistics 36. 111–152. Haegeman, Liliane and Barbara Ürögdi. 2010b. Operator movement, referentiality and intervention. Theoretical Linguistics 36. 233–246. Hartmann, Jutta M. 2005. Wh-movement and the small clause analyses of the English thereconstruction. Leiden Papers in Linguistics 2. 93–106. Harwood, Will. 2012. There are several positions available: English intermediate subject positions. To appear in Proceedings of ConSOLE XIX. Hawkins, John A. 2004. Efficiency and complexity in grammars. Oxford: Oxford University Press. Heck, Fabian. 2009. On certain properties of pied-piping. Linguistic Inquiry 40. 75–111. Heck, Fabian, Gereon Müller and Jochen Trommer. 2008. A phase-based approach to Scandinavian definiteness marking. In Charles B. Chang and Hannah J. Haynie (eds.), WCCFL 26. 226–233. Somerville, MA: Cascadilla Proceedings Project. Heck, Fabian and Malte Zimmermann. 2004. DPs as phases. Ms., Universität Leipzig and Humboldt-Universität zu Berlin. Hegarty, Michael. 1990. On adjunct extraction from complements. In Lisa Cheng and Hamida Demirdash (eds.), Papers on wh-movement (MIT Working Papers in Linguistics 13), 101–124. Cambridge, MA: MIT. Hiramatsu, Kazuko. 1999. Subject and adjunct island asymmetries: Evidence from syntactic satiation. In Sonya Bird, Andrew Carnie, Jason D. Haugen and Peter Norquest (eds.), Proceedings of the West Coast Conference on Formal Linguistics 18. 183–192. Somerville, MA: Cascadilla Press. Hiramatsu, Kazuko. 2000. Accessing linguistic competence: Evidence from children’s and adults’ acceptability judgments. Storrs, CT: University of Connecticut dissertation. Hofmeister, Philip. 2007. Retrievability and gradience in filler-gap dependencies. In Proceedings of the Chicago Linguistic Society 43. 109–123. Hofmeister, Philip. 2008. Representational complexity and memory retrieval in language comprehension. Stanford, CA: Stanford University dissertation. Hofmeister, Philip. 2009. Memory Retrieval Effects on Filler-Gap Processing. In Danielle S. McNamara and J. Greg Trafton (eds.), Proceedings of the 29th Annual Cognitive Science Society. 1091–1096. Austin, TX: Cognitive Science Society. Hofmeister, Philip and Ivan A. Sag. 2010. Cognitive constraints and island effects. Language 86. 366–415. Hofmeister, Philip, T. Florian Jaeger, Inbal Arnon, Ivan A. Sag, and Neal Snider. 2007. Locality and accessibility in wh-questions. In Sam Featherston and Wolfgang Sternefeld (eds.), Roots: Linguistics in search of its evidential base. 185–206. Berlin: Mouton de Gruyter. Hofmeister, Philip, T. Florian Jaeger, Inbal Arnon, Ivan A. Sag, and Neal Snider. In press. The source ambiguity problem: Distinguishing the effects of grammar and processing on acceptability. Language and Cognitive Processes. Hofmeister, Philip. 2011. Representational complexity and memory retrieval in language comprehension. Language and Cognitive Processes 26. 376–405. Hofmeister, Philip. 2012. Effects of processing on the acceptability of “frozen” extraposed constituents. Ms., University of Essex. Horn, George M. 1974. The noun phrase constraint. Amherst, MA: University of Massachussets dissertation. Hornstein, Norbert. 1977. S and the X′ convention. Linguistic Analysis 3. 137–176. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM 144 Liliane Haegeman et al. Huang, Cheng-Teh James. 1982a. Logical relations in Chinese and the theory of grammar. Cambridge, MA: MIT dissertation. Huang, Cheng-Teh James. 1982b. Move wh in a language without wh-movement. The Linguistic Review 1. 369–416. Ihsane, Tabea. 2008. The layered DP: Form and meaning of French indefinites. Amsterdam: JohnBenjamins. Ishii, Toru. 2009. Amelioration of the Subject Condition Effects by Remnant Movement. Paper presented at the 4th International Conference on Formal Linguistics, Beijing Foreign Studies University, July 2009. Jäger, Gerhard and Anette Rosenbach. 2006. The winner takes it all – almost: Cumulativity in grammatical variation. Linguistics 44. 937–971. Jaworska, Ewa. 1986. Prepositional phrases as subjects and objects. Journal of Linguistics 22. 355–375. Jiménez-Fernández, Ángel L. 2009. On the composite nature of subject islands: A phase-based approach. SKY Journal of Linguistics 22. 91–138. Jiménez-Fernández, Ángel L. 2012. A new look at subject islands: The phasehood of definiteness. Anglica Wratislaviensia 50. 137–168. Jurka, Johannes. 2010. The importance of being a complement: CED effects revisited. College Park, MD; University of Maryland dissertation. Jurka, Johannes, Chizuru Nakao and Akira Omaki. 2011. It’s not the end of the CED as we know it: Revisiting German and Japanese subject islands. In Mary Byram Washburn et al. (eds.), Proceedings of the 28th West Coast Conference on Formal Linguistics, 124–132. Somerville, MA: Cascadilla Proceedings Project. Kayne, Richard S. 1981. ECP extensions. Linguistic Inquiry 12. 93–133. Kayne, Richard S. 2002. Some prepositions that look DP-internal: English of and French de,” Catalan Journal of Linguistics 1. 71–115. Keller, Frank. 1998. Gradient grammaticality as an effect of selective constraint re-ranking. Papers from the 34th meeting of the Chicago Linguistic Society 2, 95–109. Keller, Frank. 2000. Gradience in grammar. Experimental and computational aspects of degrees of grammaticality. Edinburgh: University of Edinburgh dissertation. Keller, Frank. 2005. Linear Optimality Theory as a model of gradience in grammar. In Gisbert Fanselow, Caroline Féry, Ralph Vogel and Matthias Schlesewsky (eds.), Gradience in grammar: Generative perspectives, 270–287. Oxford: Oxford University Press. Keller, Frank and Antonella Sorace. 2003. Gradient auxiliary selection and impersonal passivization in German: An experimental investigation. Journal of Linguistics 39. 57–108. Kitagawa, Yoshihisa. 1986. Subjects in English and Japanese. Amherst, MA: University of Massachusetts dissertation. Kluender, Robert. 1998. On the distinction between strong islands and weak islands: Aprocessing perspective. In Peter W. Culicover and Louise McNally (eds.), The limits ofsyntax (Syntax and Semantics 29), 241–279. New York: Academic Press. Kluender, Robert. 2004. Are subject islands subject to a processing account? In Vineeta Chand, Ann Kelleher, Angelo J. Rodríguez and Benjamin Schmeiser (eds.), West Coast Conference on Formal Linguistics (WCCFL) 23, 101–125. Somerville, MA: Cascadilla Press. Kluender, Robert and Marta Kutas. 1993. Subjacency as a processing phenomenon. Language and Cognitive Processes 8. 573–633. Koopman, Hilda and Dominique Sportiche. 1991. The position of subjects. Lingua 85. 211–258. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM Deconstructing the Subject Condition 145 Koster, Jan. 1978. Why subject sentences don’t exist. In Samuel J. Keyser (ed.), Recent transformational studies in European languages, 53–64. Cambridge, MA: MIT Press. Kratzer, Angelika. 1996. Severing the external argument from its verb. In Johan Rooryck andLaurie Zaring (eds.), Phrase structure and the lexicon, 109–137. Dordrecht: Kluwer. Kuno, Susumu. 1973. Constraints on internal clauses and sentential subjects. Linguistic Inquiry 4. 363–385. Kuno, Susumu. 1987. Functional syntax – anaphora, discourse and empathy. Chicago, IL and London: The University of Chicago Press. Kuroda, Sige-Yuki. 1988. Whether we agree or not. Lingvisticae Investigationes 12. 1–47. Lasnik, Howard. 2001a. When can you save a structure by destroying it? NELS 31. 301–320. Lasnik, Howard. 2001b. Subjects, objects, and the EPP. In William Davies and Stanley Dubinsky (eds.), Objects and other subjects, 103–121. Dordrecht: Kluwer. Lasnik, Howard and Myung-Kwan Park. 2003. The EPP and the subject condition under sluicing. Linguistic Inquiry 34. 649–660. Lasnik, Howard and Mamoru Saito. 1992. Move alpha: Conditions on its application and output. Cambridge MA: MIT Press. Lebeaux, David. 1988. Language acquisition and the form of the grammar. Amherst, MA: University of Massachusetts dissertation. Legate, Julie Anne. 2003. Some interface properties of the phase. Linguistic Inquiry 34. 506–515. Levin, Beth, and Rappaport-Hovav, Malka. 1995. Unaccusativity: At the syntax-lexical semantics interface. Cambridge, MA: MIT Press. Lohndal, Terje. 2011. Freezing effects and objects. Journal of Linguistics 47. 163–199. Lyons, Christopher. 1990. An agreement approach to clitic doubling. Transactions of the Philological Society 88. 1–57. Manzini, Rita. 1992. Locality. Cambridge, MA: MIT Press. Matushansky, Ora. 2005. Going through a phase. In Martha McGinnis and Norvin Richards (eds.), Perspectives on phases, 157–181. MIT Working Papers in Linguistics 49. Cambridge, MA: MIT Press. Maeda, Masako. 2010. The criterial freezing effect and split A′-movement. English Linguistics 27. 270–296. Mahajan, Annop. 1992. The Specificity Condition and the CED. Linguistic Inquiry 23. 510–516. May, Robert. 1977. The grammar of quantification. Cambridge, MA: MIT dissertation. Martí, Luisa. 1999. Extraction from subjects in Spanish. Ms., University of Connecticut, Storrs. McCloskey, James. 1997. Subjecthood and subject positions. In Liliane Haegeman (ed.), Elements of grammar, 197–235. Dordrecht: Kluwer. McCloskey, James. 2002. Resumption, successive cyclicity, and the locality of operations. InSamuel David Epstein and T. Daniel Seely (eds.), Derivation and explanation in the Minimalist Program, 184–226. Oxford: Blackwell. Merchant, Jason. 1999. The Syntax of Silence. Santa Cruz, CA: University of California dissertation. Merchant, Jason. 2001. The syntax of silence: Sluicing, islands, and the theory of ellipsis. Oxford: Oxford University Press. Merchant, Jason. 2002. Swiping in Germanic. In Werner Abraham and Jan-Wouter Zwart (eds.), Studies in comparative Germanic syntax, 295–321. Amsterdam: John Benjamins. Brought to you by | Biblioteca de la Universidad de Sevilla Authenticated Download Date | 3/11/15 2:03 PM