ON OBJECT DISPLACEMENT IN ENGLISH PASSIVES[1] Andrew Radford (radford@essex.ac.uk) 1. Introduction In his (1999) paper Derivation by Phase, Chomsky argues that expletive passive constructions in English (i.e. passives with an expletive there subject) involve obligatory leftward or rightward displacement of the direct object. He maintains that the relevant displacement operations are phonological rather than syntactic in nature, and discusses the criterial properties of phonological and syntactic movement operations. In this paper, I challenge Chomsky’s view that the relevant object-displacement operations are phonological in nature and argue instead that neither construction involves any objectdisplacement operation. I therefore conclude that there is no evidence (at least, from the kind of structures discussed here) for weakening the theory of grammar by positing phonological phrase-movement operations. 2. Ordinary and expletive passives English has two different types of passive construction – ordinary passives like (1a) below and expletive passives like (1b/c/d)[2]: (1)(a) (b) (c) (d) Several large packages were placed on the table There were several large packages placed on the table There were placed on the table several large packages *There were placed several large packages on the table Chomsky (1999, p.20) claims that expletive passives like (1b) are ambiguous between a verbal/dynamic passive interpretation (on which (1b) is roughly synonymous with (1c)) and an adjectival/stative passive interpretation involving an ‘existential construction “there be NP”, where NP includes a reduced relative’ (on which (1b) would be paraphraseable as ‘There were several large packages which were placed on the table’). He argues that an existential analysis is appropriate for sentences like (2a) below, but not for those like (2b) (# indicates anomaly): (2)(a) #There are many cakes expected to be baked in that oven (b) There are expected to be many cakes baked in that oven Structures like (2a), he maintains, have existential import: (2a) states that there are many cakes such that they are expected to be baked in that oven, and is odd because it presupposes that the cakes already exist prior to being baked in the oven. By contrast, true expletive passive structures like (2b) have no existential import (and so (2b) is not odd). Moreover, observes Chomsky (1999, p.20), the passive participle expression is an island for extraction in existential structures (in the same way as relative clauses are), but not in expletive passives: cf. (3)(a) (b) *Which oven are there many cakes expected to be baked in? Which oven are there expected to be many cakes baked in? Chomsky does not propose any specific analysis of existential passives, but rather concentrates on expletive passives. In both ordinary passives like (1a) and expletive passives like (1b/c/d), Chomsky posits that the italicised nominal several large packages originates as a theta-marked direct object argument of the passive participle placed. Some apparent evidence in support of this claim comes from the fact that the direct object in expletive passives can be an object idiom chunk, as we see from sentences like: (4)(a) (b) (c) (d) There was umbrage taken at his remark There was due homage paid to the former president There were close tabs kept on the X-files by the FBI There was little heed paid to his warning If (as is traditionally assumed) the italicised idiom chunk nominal can only serve as the complement of the bold-printed verb[3], it is plausible to suppose that it originates as the object of the bold-printed passive participle such cases. 3. The derivation of passives Although Chomsky makes no specific claim about the syntax of the direct object of place in sentences like (1), one possibility consistent with the VP-shell analysis which he adopts (following earlier work by Larson 1988, 1990 and Hale and Keyser 1991, 1993, 1994) is that several large packages originates in spec-VP, so that at the end of the vP cycle all four sentences in (1) would have the structure (5) below[4], if we assume that V raises to adjoin to the null light verb ø which heads vP[5]: (5) [vP [v placed+ø] [VP several large packages [V placed] on the table]] Merging the Tense auxiliary be with the resulting vP in (5) will in turn derive[6]: (6) [TP [T be] [vP [v placed+ø] [VP several large packages [V placed] on the table]] The finite tense-auxiliary be agrees in person and number with (and assigns nominative case to) the closest nominal which it c-commands, namely several large packages: in consequence, be is ultimately spelled out as the third person plural form were. T has an obligatory EPP feature which requires it to project a specifier. In simple passives like (1a), the lexical array (i.e. the set of items taken from the lexicon which are used to form the relevant structure) contains no expletive and the EPP requirement of T is satisfied by moving the direct object people into spec-TP to become the subject of were, deriving[7]: (7) [TP Several large packages [T were] [vP [v placed+ø] [VP slp [V placed] on the table]] The resulting TP in (4) is then merged with a null declarative complementiser to derive the CP structure associated with (1a) Several large packages were placed on the table. In the case of expletive passives like (1b/c/d), the lexical array contains expletive there; merging this in spec-TP in (6) will derive the structure (8) below: (8) [TP There [T were] [vP [v placed+ø] [VP several large packages [V placed] on the table]] Expletive there carries an uninterpretable person feature which is erased via agreement with were. Subsequently merging (8) with a null declarative complementiser will derive the CP structure associated with (1d) *There were placed several large packages on the table. But the resulting sentence is ungrammatical. Why should this be? Chomsky (1999, p. 15) observes that passive and unaccusative objects like those italicised in (9) below cannot remain in situ in English: (9)(a) (b) *There were placed several large packages on the table (= 1d) *There arrived several large packages in the mail He concludes (1999, p.16) that ‘English bars surface structures of the form [V-DO], where the construction is unaccusative/passive.’ (For succinctness, this is referred to below as the V-DO constraint.) Chomsky has little to say about the nature of this constraint, other than noting that it must reflect a parametrised property since there are languages like Italian and Dutch in which the counterparts of sentences like (9a/b) are grammatical (i.e. languages in which the V-DO constraint does not hold). However, Chomsky notes that structures like (9a) become grammatical in English if the direct object is displaced leftwards as in (10a) below (repeated from (1b) above) or rightwards as in (10b) (repeated from (1c) above): (10)(a) (b) There were several large packages placed on the table There were placed on the table several large packages He refers to the relevant leftward displacement operation as Thematization/TH and to its rightward counterpart as Extraction/EX. He hypothesises (1999, p.18) that leftwardmoved thematized objects move to spec-vP and that rightward-moved extracted objects adjoin to vP: in consequence of the V-DO constraint, an expletive passive direct object must obligatorily undergo either Thematization or Extraction. 4 The phonological nature of object displacement A key postulate of Chomsky’s analysis of expletive passive object displacement is that both Extraction and Thematization are phonological rather than syntactic movement operations. The two types of movement operation have in common the fact that they both leave behind a trace which is phonologically null, though differ in that phonological traces are given a null spellout at the point where phonological displacement applies, whereas syntactic traces are given a null spellout at the end of the relevant phase. Moreover, the moved constituent is visible at both the LF and the PF interfaces in the case of syntactic movement, but is visible only at the PF interface in the case of phonological movement: from this is follows that syntactic movement operations may have semantic effects (e.g. a moved DP may be interpreted as specific/definite), but phonological movement operations may not. Further differences which Chomsky postulates include the fact that phonological displacement operations apply to the output of syntactic movement operations (but not conversely), and that syntactic (but not phonological) movement operations can be iterated (i.e. can apply in a successive-cyclic fashion). Given these (and earlier) assumptions, (10a) There were several large packages placed on the table and (10b) There were placed on the table several large packages would have the common syntactic structure (11) below (if the verb placed remains in situ in the syntax[8]): (11) the [CP [C ø] [TP There [T were] [vP [v ø] [VP several large packages [V placed] on table]]]] Movement of the verb placed from V to adjoin to the affixal light verb [v ø] in the PF component will derive (12) below: (12) [CP [C ø] [TP There [T were] [vP [v placed+ø] [VP several large packages [V placed] on the table]]]] However, the resulting structure violates the V-DO constraint – as we see from the ungrammaticality of (1d) *There were placed several large packages on the table. Accordingly, in the PF component the object several large packages must obligatorily undergo either leftward Thematization (deriving (13a) below), or rightward Extraction (deriving (13b) below)[9]: (13)(a) [CP [C ø] [TP There [T were] [vP several large packages [v placed+ø] [VP slp [V placed] on the table]]]] (b) [CP [C ø] [TP There [T were] [vP [v placed+ø] [VP slp [V placed] on the table] several large packages]]] (11) is the structure visible at the LF-interface, (13a/b) the structure visible at the PF interface. Chomsky offers a number of arguments in support of his claim that Thematization and Extraction are phonological operations. One type of argument which he adduces is semantic in nature. He notes (1999, p.16) that Thematization/Extraction differ from typical syntactic displacement rules ‘in not yielding the usual surface-semantic effects (specificity, etc.)’: this would follow if principles of LF-interpretation can only ‘see’ syntactic structures like (12) and not phonological structures like (13a/b). Moreover, it would then follow that ‘The English constructions reach LF in the same form as in similar languages, as we would expect if LF-external systems of interpretation are essentially language-independent and prefer the LF interface to be as uniform as possible across languages’ (Chomsky, 1999, p. 16). Chomsky goes on to note that his analysis of Thematization and Extraction as phonological operations correctly predicts that (in English expletive passives) the whole direct object is inaccessible to syntactic movement operations like wh-movement – as we see from the ungrammaticality of expletive passives such as[10]: (14) *How many packages were there placed on the table? (14) cannot be derive from a syntactic structure with an in situ direct object, since (as (13) above shows) by the time wh-movement applies on the CP cycle, the direct object will already have obligatorily undergone either Thematization or Extraction in the phonological component on the vP cycle[11], leaving behind a null trace which cannot undergo movement if ‘traces are inaccessible to Move’ (Chomsky 1999, p.18). Nor can (14) be derived from phonological structures like (13a/b), since phonological structures are inaccessible to syntactic movement operations. Chomsky also argues that his analysis of Thematization and Extraposition as phonological operations correctly predicts that no subpart of the direct object can undergo syntactic movement. In this connection, consider the contrast below[12]: (15)(a) (b) (c) ?How many people did they place pictures of on the table? *How many people were there pictures of placed on the table? *How many people were there placed on the table pictures of? The slight awkwardness of sentences like (15a) may be ‘the result of incomplete internal constituent constraints on movement (Kuno 1973)’ (Chomsky 1999, fn 41). However, (15b/c) are completely ungrammatical. This, Chomsky argues, is because (15b/c) could only be derived by application of wh-movement to phonological structures like (13a/b); but it is in the nature of phonological movement operations that their output in inaccessible to syntactic movement operations. Chomsky argues that Thematization and Extraction also differ from syntactic operations in that they cannot be iterated (i.e. they cannot apply to their own output). Data such as (16) below might be seen as providing empirical support for this claim in relation to Thematization: (16)(a) There had been umbrage taken at his remarks (b) *There had umbrage been taken at his remarks Chomsky does not, however, explain why phonological movement operation should not be iterable, simply noting (1999, p.19) that non-iterability is ‘perhaps a more general property of operations not driven by uninterpretable features, and/or phonological operations[13].’ As should be apparent, Chomsky’s analysis of the syntax of expletive passive direct objects in English is of considerable theoretical importance because it leads him to the conclusion that natural language grammars incorporate both syntactic and phonological XP-movement operations. However, from a conceptual perspective, his postulation of phonological movement operations can be seen as introducing an unwanted imperfection into the theory of grammar. Within a perfect theory of grammar, we might conjecture, the syntactic (i.e. computational) component of the grammar would generate a single structural representation which is visible at both the LF and the PF interfaces. However, within the model outlined in Chomsky (1999), the syntactic component generates a syntactic structure (visible only at the LF interface) which is mapped into a distinct phonological structure (visible only at the PF interface). In other words, Chomsky’s analysis in some ways marks a retreat to earlier two-level models of syntax in which a distinction was drawn between deep structure/shallow structure and surface structure. Of course, if (as claimed here) there are no phonological object displacement operations in English expletive passives (and, perhaps more generally, no phonological movement operations[14]), we can maintain a single-level theory of structural representation. In the sections which follow below, I argue against positing either of the phonological object displacement operations which Chomsky postulates. 5. On the non-existence of Extraction There are a number of theoretical and descriptive problems posed by Chomsky’s analysis of clause-final objects in expletive passives like (1c) There were placed on the table several large packages. One relates to the question of what drives (rightward) Extraction of the object in such cases. The only answer which Chomsky offers to this question (1999, p.18) is to suggest the possibility that ‘a weak phase has a phonological counterpart to EPP’: in other words, the head light verb of a passive vP may carry some kind of EPP-feature which triggers phonological movement. But since a head H with an EPP feature typically triggers (leftward) movement of a phrasal constituent to specHP[15], it is not clear how an EPP-feature of a light verb v could drive (rightward) adjunction to vP. What makes Chomsky’s analysis even more problematic is the fact that he has to posit that the clause-final direct object in sentences such as: (17)(a) There are likely to be awarded several prizes (b) We expect there to be awarded several prizes undergoes string-vacuous Extraction and is thereby adjoined to the righthand edge of vP. Since movement is a prima facie imperfection which is tolerated only to the extent that it resolves interface problems, the question arises of what specific PF-interface problem vacuous Extraction in the PF component is designed to resolve. Clearly, it cannot be linear ordering problems, since Extraction in such cases does not change the linear order of the overt constituents in (16). Moreover, Chomsky’s Extraction analysis violates Kayne’s (1994) Linear Correspondence Axiom/LCA under which x precedes y iff x asymmetrically c-commands y. Chomsky (1995, 334-340) adopts LCA (while modifying some of Kayne’s assumptions) and concludes that it follows from LCA that in all structures in which an XP is adjoined to some target constituent, ‘the adjunct precedes the target’. However, under Chomsky’s analysis of Extraction, a clause-final direct object in an expletive passive undergoes rightward adjunction to vP, in violation of LCA. Given the problems posed by Chomsky’s Extraction analysis, I shall explore an alternative analysis under which clause-final expletive passive direct objects are in situ constituents which undergo no movement operation of any kind. On this view, the direct object several large packages in a sentence like (1c) There were placed on the table several large packages is merged as the complement of placed and the locative argument on the table as the specifier of placed, so that (1c) has the structure (18) below[16]: (18) [CP [C ø] [TP There [T were] [vP [v placed+ø] [VP on the table [V placed] several large packages]]]] The object several large packages would remain in situ. Since the analysis proposed here does not assume any rightward movement, the need to attribute the non-iterability of rightward movement of direct objects to a (somewhat mysterious) property of phonological operations does not arise. Moreover, the analysis sketched here provides a straightforward account of the ungrammaticality of sentences like (1d) *There were placed several large packages on the table. This follows from the assumption that direct objects are always merged (and theta-marked) in VP-complement position. A sentence like (1d) would require a structure like (19) below: (19) [CP [C ø] [TP There [T were] [vP [v placed+ø] [VP several large packages [V placed] on the table]]]] But (19) would be ungrammatical by virtue of violating the requirement for direct objects to be merged in VP-complement position. Thus, the analysis proposed here obviates the need to posit a V-DO constraint which filters out unaccusative or passive V-DO structures (a constraint which amounts to a return to the kind of output filters associated with Chomsky and Lasnik’s (1977) Filters and Control paper – a device criticised in subsequent work for being entirely ad hoc). The in situ analysis of expletive passive direct objects would also enable us to avoid the need for positing vacuous rightward movement of the direct object in sentences like: (20) There were awarded several prizes Under the in situ analysis proposed here, (20) would have the structure (21) below: (21) [CP [C ø] [TP There [T were] [vP [v awarded+ø] [VP [V awarded] several prizes]]]] The clause-final direct object several prizes would originate in VP-complement position, in accordance with our earlier assumption that direct objects always originate as the complement of V[17]. The object several prizes would not undergo vacuous rightward movement, but rather would remain in situ in VP-complement position. Under the in situ analysis of clause-final expletive passive direct objects sketched in this section, the direct object is first-merged as the ‘lowest’ argument of the verb. This in turn means that other internal arguments of the verb will c-command the direct object. Some evidence that this is indeed the case is provided by sentence such as (22) below: (22) There were shown to the blackmail victims numerous compromising photos of themselves Under the analysis proposed here, the italicised string in (22) will have the syntactic structure indicated in (23) below[18]: (23) [VP to the blackmail victims [V' [V shown] numerous compromising photos of themselves]] and the oblique argument to the blackmail victims will c-command and hence be able to bind the reflexive themselves[19]. Further evidence that oblique arguments originate ‘above’ direct objects come from contrasts such as: (24)(a) (b) It was reported to the authorities that several prisoners had escaped *It was reported that several prisoners had escaped to the authorities In general, that​-clause direct objects typically remain in situ and do not move to some higher position within vP/VP (arguably because they are case-resistant, as noted by Stowell 1981)[20]. If so, the fact that the that-clause direct object in (24) must precede the indirect object to the authorities suggests that VP-complement position is the canonical position for first-merge of direct objects[21]. It may be that an in situ direct object is typically assigned a specific type of interpretation (e.g. as focused or new or contrastive), and that heavy objects are the optimal candidates for the relevant interpretation, thereby accounting for the heaviness effect. What remains to be accounted for under the in situ analysis proposed here are extraction facts. The grammaticality of sentences such as: (25) How many people were there arrested? would suggest that there is no constraint against the whole object undergoing a syntactic displacement operation like wh-movement. Moving a subpart of the object, however, yields ungrammaticality, as we saw earlier in relation to (15)(c) above – repeated as (26) below: (26) *How many people were there placed on the table pictures of? However, we find a similar picture in relation to in situ object complements of a VP containing a prepositional specifier, as contrasts such as the following illustrate: (27)(a) (b) He sent to you which copy of the book? (echo question) Which copy of the book did he send to you? (28)(a) (b) He sent to you a copy of which book? (echo question) *Which book did he send to you a copy of? For reasons which are not clear, it would seem that extracting a subpart of an in-situ VPcomplement across an oblique argument leads to ungrammaticality. Whatever the precise nature of the relevant constraint[22], it provides us with an account of the ungrammaticality of sentences like (26). 6. On the non-existence of Thematization Having looked at Chomsky’s analysis of rightward extraction, I now turn to look at his analysis of leftward Thematization of objects in sentences like (1b) There were several large packages placed on the table. There are a number of questions which arise about the descriptive adequacy of the claims made by Chomsky. For one thing, his claim that Thematization cannot apply in a successive-cyclic fashion would seem to be called into question by sentences such as: (29)(a) (b) There are continually being new treatments developed for cancer There are continually new treatments being developed for cancer (30)(a) (b) He could see that there was being umbrage taken at his remarks He could see that there was umbrage being taken at his remarks (29a/30a) is the output we obtain (under Chomsky’s analysis) if the thematized object moves to spec-vP (where it is supposedly is frozen in place); (29b/30b) is the output we obtain if the italicised object moves in successive-cyclic fashion from spec-vP to become the specifier of the superordinate projection headed by being. The grammaticality of (29b/30b) might at first sight seem to lead us to the conclusion that Thematization (contrary to what Chomsky suggests) can indeed be iterated and hence arguably is a syntactic operation akin to (iterable) A-movement. However, the conclusion that Thematization is a syntactic A-movement operation and hence can be iterated in a successive-cyclic fashion raises the question of why iteration is not possible in structures such as (16) above, repeated as (31) below: (31)(a) (b) There had been umbrage taken at his remarks *There had umbrage been taken at his remarks Any attempt to handle the difference in grammaticality between (30b) and (31b) by positing that being has an EPP-feature (requiring it to project a specifier) but been does not would not only be purely stipulative in character but would also fail to account for the ungrammaticality of the (b) examples in (32-33) below: (32)(a) There ended up being several demonstrators arrested (b) *There ended up several demonstrators being arrested (c) Several demonstrators ended up being arrested (33)(a) There kept being complaints made about the noise (b) *There kept complaints being made about the noise (c) Complaints kept being made about the noise The ungrammaticality of sentences like (32b/33b) calls into question the plausibility of a successive-cyclic A-movement analysis. A closer look at the data in (29-33) suggests the generalisation that in English expletive passives, the associate must occupy a surface position immediately below be (in the sense that be must be the closest verb c-commanding the associate). Hence, we have grammatical outcomes in (29a) with new treatments following being, in (29b) with new treatments following are, in (30a) with umbrage following being, in (30b) with umbrage following was, in (31b) with umbrage following been, in (32a) with several demonstrators following being and in (32b) with complaints following being. Conversely, we have ungrammatical outcomes in (31b) where umbrage follows had, in (32b) where several demonstrators follows ended up, and in (33b) where complaints follows kept. Data such as (34-36) below lend further empirical support to the generalisation that a preparticipial associate in expletive passives must be positioned below be: (34)(a) (b) Several demonstrators got arrested by the police *There got several demonstrators arrested by the police (35)(a) (b) They wanted replica guns issued to the campus cops *They wanted there issued to the campus cops replica guns (36)(a) (b) (c) There seem to have been several passengers injured *There seem to have several passengers been injured *There seem several passengers to have been injured The most straightforward conclusion to arrive at on the basis of data such as (29-36) is that the relevant sentences are existential structures of the form there+be+associate+participle (with the participial expression modifying the associate in some way) – contrary to what Chomsky claims. An existential analysis would also resolve a number of other puzzles which surround Chomsky’s Thematization analysis. One is why the verb be is apparently immune from the V-DO constraint, as the following contrast shows: (37)(a) (b) There were several large packages on the table *There were placed several large packages on the table In terms of the assumptions made here, we can provide a straightforward answer to this question by positing that structures like (37a) are existential be structures in which the closest verb c-commanding the italicised associate is be[23], whereas in (37b) the closest verb c-commanding the italicised associated is placed. The existential analysis proposed here also provides the key to a further mystery in Chomsky’s analysis. A key claim which Chomsky makes is that not only passive participles but also unaccusative verbs are subject to the V-DO constraint – a claim illustrated in terms of contrasts such as the following: (38)(a) (b) *There came several angry men into the room There came into the room several angry men (39)(a) (b) *There arrived a strange package in the mail There arrived in the mail a strange package Under Chomsky’s analysis, structures like (38a/39a) are barred by the V-DO constraint, and structures such as (39b/39b) are the result of rightward Extraction of the italicised object[24]. But given the assumption that intransitive objects can undergo leftward Thematization in passive participle structures, we should expect the same leftward movement of the object to be possible with unaccusative participles. However, it is not, as the ungrammaticality of examples like those in (40) below illustrates: (40)(a) (b) *There had several angry men come into the room *There has a strange object arrived in the mail Nothing in Chomsky’s analysis (as far as I can see) precludes leftward Thematization from applying to unaccusative objects in the same way as it applies to passive objects; and simply stipulating that passive participles have an EPP-feature which unaccusative participles lack would be an ad hoc expedient entirely lacking in explanatory force. By contrast, under the existential analysis of There+be+associate+participle structures proposed here, the ungrammaticality of structures like (40a) can be accounted for straightforwardly by positing that the relevant structures lack the existential predicate be which is required in the relevant existential expletive construction in English. 7. Conclusion The main goal of this paper has been to argue against the existence of the two phonological XP-movement operations which Chomsky posits in his analysis of expletive passive/unaccusative structures in English. I argued that the clause-final object in expletive passives like There were placed on the table several large packages is an insitu object which is first-merged in the canonical VP-complement position associated with objects. Likewise, I argued for a parallel conclusion in relation to clause-final objects in expletive unaccusatives such as There arrived in the mail a strange package. I also argued that sentences such as There were several large packages placed on the table are existential structures of the form there+be+associate +participial modifier, in which the associate is likewise generated in situ. Under the analysis proposed here, a sentence like How many large packages were there placed on the table? could in principle either be the wh-movement counterpart of an existential structure such as There were several large packages placed on the table or the wh-counterpart of a genuine expletive passive structure such as There were placed on the table several large packages[25]. 8. References Bowers, J. (1993) ‘The syntax of predication’, Linguistic Inquiry 24: 591-656. Chomsky, N (1999) Derivation by Phase, MIT Working Papers in Linguistics Chomsky, N & Lasnik H (1977) ‘Filters and Control’, Linguistic Inquiry 8: 425 -504 Hale, K & Keyser, S J (1991) On the Syntax of Argument Structure, Lexicon Project Working Papers, Center for Cognitive Science, MIT Hale, K & Keyser, S J (1993) `On argument structure and the lexical expression of semantic relations’, in K Hale & S J Keyser (eds) The View from Building 20, MIT Press, Cambridge Mass, pp.53-109 Hale, K & Keyser, S J (1994) ‘Constraints on argument structure’, in B Lust, M Suñer & J Whitman (eds) Heads, Projections and Learnability, Erlbaum, Hillsdale NJ, vol.1, pp.53-71 Kayne, R (1994) The Antisymmetry of Syntax, MIT Press, Cambridge Mass. Kuno, S (1973) ‘Constraints on Internal Clauses and Sentential Subjects’, Linguistic Inquiry 4: 363-385 Larson, R (1988) ‘On the double object construction’, Linguistic Inquiry 19: 335 -391 Larson, R (1990) ‘Promise and the theory of control’, Linguistic Inquiry 22: 103-39 Pesetsky, D & Torrego E (2000) ‘T-to-C movement: Causes and consequences’, to appear in M Kenstowicz (ed) Ken Hale: A Life in Language, MIT Press, Cambridge Mass. Stowell, T (1981) Origins of Phrase Structure, PhD diss, MIT ___________________________________________________________________ [1] I am grateful to members of the LAUG Research Group at Essex (Martin Atkinson, Bob Borsley, Claudia Felser, Mike Jones and Roger Hawkins) for helpful comments on an earlier draft of this paper. This version appeared in Essex Research Reports in Linguistics (November 2000), vol. 33, pp.33-49. [2] Impersonal passives like It was agreed that the chairman should be reprimanded are another type of passive with expletive subjects, but are not discussed here since they do not manifest the phenomenon of object displacement which is the central focus of this paper. [3] As Mike Jones points out, sentences such as Any umbrage which was taken at his remarks was soon forgotten or The homage which was paid to the former president was gratifying might suggest that this claim must be weakened somewhat, since a relative pronoun with an idiom chunk antecedent can serve as the object of the relevant verb. [4] Throughout this paper, labelled bracketings are simplified by showing only heads and maximal projections, and by not showing structure not immediately relevant to the discussion at hand: material which is deleted at the end of the relevant phase is indicated by strikethrough and traces are shown in a smaller typeface. (1b) only has the structure shown here at the relevant stage of derivation as an expletive passive, not as an existential passive. [5] If Bowers (1993) is correct in supposing that perfectly is a VP-adverb, the position of the participle manoeuvred in The shuttle was manoeuvred perfectly into position would suggest that the participle raises from V to v. To simplify exposition, I overlook the suggestion made by Chomsky that some or all head-movement operations may be phonological in nature: it is not clear whether Chomsky would hold that this is the case for movement from V to v. [6] To simplify exposition, I am assuming here that be originates in T, though the fact that passive be has nonfinite forms may suggest that it originates in a lower position and from there raises to T. However, this is an incidental detail which does not affect the main thrust of the argumentation here. [7] Because of space constraints, the trace of several large packages is represented as slp [8] I am assuming here that V-movement is a phonological operation; but it does not affect the argumentation here if it is a syntactic operation. [9] To save space, the trace of several large packages is shown as slp. Recall that labelled bracketings are simplified by not showing intermediate projections – only heads and maximal projections. [10] (12) is grammatical as an existential (adjectival) passive, but this is irrelevant to Chomsky’s point. Personally, I also find it grammatical as an expletive (verbal) passive as well – i.e. for me it is ambiguous. [11] It is not clear what makes Thematization/Extraction obligatory in Chomsky’s analysis – as noted later in the text. [12] (13b) is more acceptable as an existential/adjectival passive, but this is irrelevant here [13] It is not clear to me why Chomsky posits that phonological operations cannot be iterated. If (as he suggests) head movement may be a phonological operation, it would appear that it can be iterated in that (e.g.) a finite verb can move to T and thence to C in successive (iterated) steps in (British English) structures such as Have you any wool? [14] Other, perhaps, than phonological cliticisation operations by which a clitic C attaches to a host H is there is no phonologically overt material intervening between the two at the stage of derivation when PF-cliticisation applies. A case in point might be auxiliary contraction structures such as Where’s he going? However, an alternative no-movement approach to PF cliticisation would be to explore the possibility suggested by Chomsky’s (1999, p.11) that phonological rules typically ‘blur or remove boundaries between units’ (which raises the possibility that contraction might result from removal of the boundaries between clitic and host where there is no overt phonological material between the two). [15] Or adjunction of a head to H, if an EPP feature can also drive head movement as in Pesetsky and Torrego (2000). [16] If V-to-v movement is phonological, (18) will be the structure at PF; if it is syntactic, (18) will be the structure at both the PF and LF interfaces. [17] By contrast, under the analysis sketched earlier, the direct object would originate in VP-specifier position in (1c) but in VPcomplement position in (17a), presupposing a non-uniform account of the syntax of direct objects. [18] If V-to-v movement is phonological [19] The grammaticality of structures such as You must talk to Mary about herself suggests that an oblique argument can ccommand and bind a reflexive. [20] One well-known exception is that that-clauses can seemingly passivise in sentences such as That several prisoners had escaped was reported to the authorities. It may be that apparent passivisation here is a topicalisation or focalisation operation of some kind. [21] This of course does not preclude the possibility that transitive direct objects move to a superordinate specifier position, e.g. as in We showed compromising pictures of themselves to the blackmail victims. [22] An interesting possibility to explore might be an incompatibility between the interpretive consequences of leaving the object in situ in VP-complement position and the interpretive consequences of extracting part of it via wh-movement. [23] Other non-verbal material can intervene between be and the associate, e.g. an adverb like definitely in a sentence such as There were definitely several large packages on the table. [24] Under the very different analysis proposed here, the structures in (38b/39b) are the result of first-merging the unaccusative object in its canonical VP-complement position, not of movement. [25] In my judgement, the relevant wh-question is ambiguous, as expected under the analysis proposed here.