Verb Phrase Types and the Notion of a Phase∗ Julie Anne Legate December 15, 1998 1 Introduction This squib investigates the notion of a phase (Chomsky 1998a,b) with respect to the categories vP and VP. Chomsky (1998a,b) introduced the notion of a phase as a self-contained subsection of a derivation defined by the category v or C. According to this proposal, after construction of a phrase marker containing one of these categories (i.e. after construction of a vP or CP), spellout occurs, sending the phase to interpretation at the PF and LF components. At this point, the derivation may terminate, or an additional lexical subarray may be chosen such that the derivation continues to build upon the previous phase. However, subsequent operations are limited in their referal to elements within the previous phase by the Impenetrability Condition. This condition allows visibility only within the edge of the phase, that is any adjuncts or specifiers of the highest v/C and the highest v/C itself. Thus, a version of strict cyclicity is built into the computation. Chomsky’s proposal explicitly makes a sharp distinction between unergative verbs, on the one hand, and passive and unaccusative verbs, on the other, since the former are presumed to consist of a VP dominated by a vP, while the latter are presumed to consist solely of a VP. Thus, the unergative verb phrase would form a phase, while the passive or unaccusative verb phrase would not. It is this distinction that we will question today. The goal of this paper is thus not to question the existence of a phase at the level of the verb phrase,1 but rather to demonstrate that the tests supporting the notion of a vP phase do not make a crucial split between unergative verbs on the one hand and passive or unaccusative verbs on the other. The conclusion reached will be if there are vP level phases then either there are also VP level phases, or passive and unaccusative verb phrases are also dominated by a vP, albeit one that lacks an external argument. ∗I would like to thank Charles Yang for comments and discussion. some have claimed that this should also be done (Pesetsky 1998, for example). 1 Although 1 2 Tests for Phase-hood I: PF The tests for phase-hood suggested by Chomsky (1998a,b) are all intended to identify some type of independence at the interfaces. Since a phase is sent to the PF and LF interfaces as a single unit, it is expected that the interfaces treat them as an independent unit, in some manner or other. At the PF interface, the possible tests cited are extraposition, (pseudo)clefting, islolation, and phrasal stress assignment through the Nuclear Stress Rule. Bresnan (1972) discusses the Nuclear Stress Rule, arguing on the basis of contrasts like those in (1) that it applies cyclically. (1) Cyclic Application of Nuclear Stress Rule 1 a. Mary liked the proposal that George leave. b. Mary liked the proposal that George left. c. George has plans to leave. d. George has plans to leave. 1 1 1 (Bresnan 1972:75) In (1a) and (1c) we see the normal application of the Nuclear Stress Rule, which assigns primary phrasal stress to the rightmost element in the VP capable of bearing it (indefinites, pronouns, and prepositions are not capable of bearing primary stress, as discussed below). In (1b) and (1d), on the other hand, the stress appears on an apparently non-VP-final constituent. Bresnan’s intuition was that the Nuclear Stress Rule applies normally in these examples, but that its application is always cyclic. Thus, assuming that “proposal” in (1b) and “plans” in (1d) are moved from the object position of the embedded clause,2 these elements receive primary phrasal stress on the first application of the Nuclear Stress Rule, before they have moved from object position: (2) The First Cycle of Nuclear Stress Assignment 1 a. George leave b. George left the proposal c. PRO to leave d. PRO to leave plans 1 1 1 2 A controversial assumption at the time, but one recently brought into favour by Kayne 1994. 2 If true, this analysis supports the notion of a phase in that the phonology appears to operate on a syntactic structure that does not exist after completion of the computation. Thus, spellout at intermediate steps in the derivation seems necessary.3 Bresnan formulates the domains of application for the Nuclear Stress Rule as S and NP, following Chomsky’s (1970) claim that transformations apply cyclicly to each of these categories. Updating her results to the notion of phase, we discover that there is an argument for an application of the Nuclear Stress Rule at the level of the verb phrase, assuming, as Chomsky (1998a,b) argues, that TP is not a phase. Consider (3). (3) vP Application of the Nuclear Stress Rule 1 a. The parable shows what (suffering men) can create. b. The parable shows (what suffering) men can create. (Bresnan 1972:79) 1 In (3a), “what” is the object of the embedded verb “create”. Bresnan shows that indefinites like “what” cannot bear primary phrasal stress even when final in the verb phrase. Instead, the primary stress lands on the rightmost element which can bear the stress. In the case of (3a), this is the verb “create”. In (3b), on the other hand, the object of the embedded verb is “what suffering”. Since “suffering” is capable of bearing primary stress, stress assigned to it. When the wh-phrase is moved to [spec, C] on the subsequent phase, “suffering” carries its primary stress with it. Thus, the Nuclear Stress Rule seems to treat the vP as a domain of operation, i.e. a phase. This pattern can be duplicated with passive and unaccusative verb phrases. Consider (4). (4) VP Application of the Nuclear Stress Rule 1 a. This story shows what can be created by suffering men. b. This story shows what suffering can be created by men. c. This story shows what suffering can ensue. 1 1 (4a) shows the application of the Nuclear Stress Rule to the verb phrase final “created”.4 In (4b), “suffering” receives the primary phrasal stress, demonstrating that the Nuclear Stress Rule applied while the wh-phrase “what suffering” was still final in the passive verb phrase headed by “created”. Similarly, in 3 The alert reader may have noticed an inconsistency between this proposal and the Impenetrability Condition. The issue will be discussed below. 4 Interestingly, the by-phrase seems invisible to the Nuclear Stress Rule. Clearly there is additional complexity here to be investigated. 3 (4c), “suffering” again receives primary phrasal stress, demonstrating that the Nuclear Stress Rule applied while it was still rightmost in the unaccusative verb phrase headed by “ensue”. Therefore, the Nuclear Stress Rule seems to operate on passive and unaccusative VPs, just as it does on unergative vPs, supporting the proposal that all three are phases. Before proceeding to the next test for phase-hood, however, an internal inconsistency must be acknowledged. The correct operation of the Nuclear Stress Rule in examples (3) and (4) requires the direct objects of these verbs to be in situ in order to be assigned primary phrasal stress as the rightmost element in the verb phrase. Chomsky’s (1998a,b) proposal of the phase, however, crucially involves the Impenetrebility Condition, mentioned in the Introduction. That is, only the edge of a previous phase is visible to the computation. This condition is crucial in that it is the primary motivation for successive cyclic whmovement targetting the vP and CP. If the wh-phrase remained in situ, the logic goes, it would not be visible to further operations in the computation, including subsequent wh-movement. Therefore, wh-phrases like “what suffering” that will be subject to further movement must move to the edge of their phase, i.e. adjoin to the vP(/CP) dominating them,5 however, in this position they will no longer be rightmost in the verb phrase and thus should not receive primary phrasal stress. I have no solution for this apparent contradiction, but simply acknowledge it as a problem. The logic of three of the other tests for phase-hood–extraposition, clefting, and pseudoclefting–is to show that the phase behaved as a unit in movement-like structures that perhaps do not involve purely syntactic movement.6 Extraposition and clefting do not apply to verb phrases, and thus cannot be used to compare the behaviour of ergative verbs with that of passive and unaccusative verbs. (5) (6) Extraposition a. It suprised John CP [that b. *It suprised John c. *It surprised John d. *It suprised John vP [saw he saw Mary in the grocery store]. Mary in the grocery store]. V P [arrive early] V P [demoted to ensign] Clefting 5 However, the motivation for the movement of a wh-phrase to the edge of its phase is assumed to be feature checking requirements, as is all movement in the model. 6 Admittedly, the analysis of extraposition, cleft, and pseudocleft, as well as the verb phrase fronting and “though” movement considered below, is contentious. More generally, it is quite certainly the case that none of the arguments for phase-hood are unassailable. However, my strategy in this squib will be to examine each possible argument in turn and show that it does not differentiate between unergative verbs and passive/unaccusative verbs. The logic is that it is much worse to ignore a potentially relevant test than to spend time on a test that later turns out to be irrelevant. 4 a. It’s DP [Mary] that John saw in the grocery store. b. *?It’s vP [see c. *?It’s V P [arrive d. *?It’s V P [demoted Mary in the grocery store] that John did. early] that John did. to ensign] that John was. Pseudoclefts, on the other hand, do target verb phrases. Consider (7). (7) vP Pseudoclefts a. What John did was [eat worms]. b. ?What John was was [teaching physics]. The example in (7a) illustrates the pseudocleft construction with a regular transitive verb phrase. Example (7b), perhaps slightly degraded, provides a pseudocleft sentence with a “be” auxiliary for comparison with the passive. (8) VP Pseudoclefts a. What John did was [arrive early]. b. ?What John was was [demoted to ensign]. As we see in (8), pseudoclefting is equally possible with unaccusative and passive verb phrases, although the passive in (8b) is perhaps less than perfect due to the “be” auxiliary. Furthermore, we may be able to use verb phrase fronting and “though” movement as similar movement-type tests that are more appropriate for verb phrases than clefting and extraposition. Consider (9). (9) vP Fronting a. John said he would eat worms, and vP [ti eat worms] hei did! b. John said he would be teaching physics (by the time he was 26), and vP [teaching physics] hei was! In (9a), the verb phrase vP [ti eat worms] has been moved as a unit to a position above TP, possibly a topic projection. Since the construction seems slightly degraded when “be” rather than “do” is the stranded auxiliary, (9b) is included for comparison with the passive verb phrase below. Now consider (10) (10) VP Fronting a. John said he would arrive early, and b. John said he would be sent to Europe, and hei was! 5 V P [arrive ti early] hei did! V P [sent ti to Europe] (10a) involves fronting of the unaccusative verb phrase V P [arrive ti early], and is perfectly grammatical. (10b) with a passive verb, on the other hand, is perhaps slightly degraded, but no more so than (9b) which also involves stranding of the auxiliary “be”. The “though” movement construction also involves fronting of a verb phrase, as shown in (11). (11) vP “though” Movement a. Teach physics though John may... b. Eating worms though John may be... (12) VP “though” Movement a. Arrive early though John may... b. Demoted to ensign though John may be... (12) illustrates that “though” movement also affects unaccusative and passive verb phrases on a par with unergative verb phrases. Thus, we have seen that unaccusatives and passives behave identically to unergatives with respect to three movement-type tests for phase-hood: pseudoclefting, verb phrase fronting, and “though” movement. Another possible test for phase-hood, isolation, also groups unaccusatives and passives with unergatives. This test is built on the assumption that convergent derivations are minimally a phase. This assumption is plausible in the model in that the derivation begins by selecting a lexical subarray for the computation. If subarrays smaller than a single phase were allowed, the concept of a phase as a useful notion would be lost. Any phrase would then be able to be computed individually and sent to PF and LF as an independent unit, eliminating the special status of the phase in this regard. Thus, if the notion of the phase is to make any predictions, we must limit the choice of subarrays to a phase, and thereby limiting the possible convergent derivations to phase-multiples. For verb phrases, the test of isolation involves so-called “Mad Magazine” sentences like those in (13). (13) vP in Isolation a. Me teach physics?! b. Me skipping class?! Although such sentences require a specific context (“John, I want you to teach physics next year”, “John, you wouldn’t be skipping class, would you?”), they are grammatical, thus supporting their characterization as a phase. The data in (14) demonstrate that this is also true of unaccusatives, (14a), and passives, (14b). 6 (14) VP in Isolation a. Me arrive early?! b. Me demoted to ensign?! (Possible contexts for these sentences are “John, for once I would like you to arrive early for class”, and “John, due to your misconduct, I am hereby demoting you to the rank of ensign”.)7 In this section, we have seen five potential arguments for the phase-hood of vP: the Nuclear Stress Rule, Pseudoclefting, verb phrase fronting, “though” movement, and isolation. All fail to distinguish between unergative verbs, on the one hand, and passive and unaccusative verbs, on the other, thus supporting the thesis that passive and unaccusative verb phrases also constitute a phase. In the next section, we consider arguments for the phase-hood of vP based on independence at LF. 3 Tests for Phase-hood II: LF Chomsky (1998a,b) suggested that the phase-hood of vP and CP is supported at LF by their status as a “proposition”, and by A-bar movement to these positions, both Quantifier Raising and successive cyclic wh-movement. I must confess to being unclear as to what is meant by “proposition” in this context that would distinguish unergative verb phrases from unaccusative and passive verb phrases. Type theory, for example, does not distinguish them, both being of type < t > (ignoring event, world, time, or other variables which both may or may not have, depending on the semantic theory assumed). Whether quantifier raising (henceforth QR) and successive cyclic wh-movement target passive and unaccusative verbs, however, can be tested. To do so, I will use antecedent contained ellipsis (henceforth ACE), parasitic gaps, and interactions between reconstruction and binding, as diagnostics. First, let’s consider antecedent contained ellipsis. ACE constructions, like that in (15), are standardly assumed to require quantifier raising, either in order 7 It may be argued that the expected form of Mad Magazine sentences built on unaccusatives and passives would involve the object following the verb, as in “*Arrive me early?!” or “*Demoted me to ensign?!”. However, there may be reason to expect the attested forms in (14) instead. Recall that the Impenetrability Condition states that only the edge of the highest phrase (i.e. its head, specifier(s), and adjunct(s)) is available to operations during a following phase. Thus, for example, in a construction involving object wh-movement, a feature on v is posited which attracts the object to the edge of vP. Thus, the object is a possible target for wh-movement to [spec, C] in the following phase. If VP is a phase, the same logic may apply to the object of passives and unaccusatives. In order for the object to be a possible target for movement to [spec, T], it would have to appear in the edge of VP. Therefore, a feature on V must attract the object to its specifier, resulting in the attested word order of the object preceding the verb. The fact that no subsequent phase is constructed, the object receiving default case, is purely incidental, and cannot be known at the point a V with the relevant feature is chosen. 7 to create parallelism between the ellided verb phrase and its antecedent, or to avoid infinite regress when the antecedent verb phrase is copied into the empty verb phrase at LF (see Bouton 1970, Sag 1976, Chomsky & Lasnik 1993, Fox 1995, Kennedy 1997, among others, for discussion of this construction). (15) Antecedent Contained Ellipsis Resolution Through Quantifier Raising a. Dawn b. Dawn likes vP 1 [ likes DP [ one of the books I do DP [ one of the books I do vP 2 [ vP 2 [ e ]]i e ]]]. vP 1 [ likes ti ] The structure in (15b) allows both the ellided verb phrase, vP2, and its antecedent, vP1, to consist of [likes ti ], thus satisfying parallelism/avoiding infinite regress. Notice, however, that all that is needed for ACE resolution is for the DP containing the ellided verb phrase to undergo quantifier raising. In order to test whether this movement can be to the verb phrase, we introduce other scope-bearing elements into the sentence. (16) Quantifier Raising to VP a. b. c. d. Mary wasn’t V P 1[ Some woman was ∃>∀ introduced to V P 1[ given The road didn’t V P 1 [ go past pected it to V P 2 [ e ]]]. Some train ∃>∀ V P 1[ arrived in DP [ DP [ DP [ DP [ anyone you were V P 2[ every message you were e ]]]. V P 2[ e ]]]. any of the scenic spots you ex- every city you expected it to V P 2[ e ]]]. In (16a) and (16c), negative polarity items were used to diagnose the position of the moved DP. In order to be licensed, the DPs must have undergone QR to a position no higher than the negation, thus to adjoin to the passive verb phrase in (16a) and to the unaccusative verb phrase in (16c).8 Similarly, in order to obtain the most salient reading of (16b) and (16d), in which the existential has scope over the universal, the DP must have undergone QR to a position below the subject, again to adjoin to the passive verb phrase in (16b) and to the unaccusative verb phrase in (16d). Thus, we conclude that quantifier raising targets passive and unaccusative verb phrases equally to unergative verb phrases. Now let’s consider the intermediate traces of wh-movement. The logic of this test is that in order for a wh-word to be visible to movement operations during a subsequent phase, it must move to the edge of its phase, in accordance with the Impenetrability Condition. Thus, if CP and vP are both phases, then 8 Assuming that the licensing of NPIs happens at LF rather than S-Structure, the latter no longer a relevant notion in the theory. See Urribe-Etxebarria (1996) for arguments that NPIs are licensed in the scope of negation at LF. 8 successive cyclic wh-movement must leave copies at every intermediate CP/vP. Lebeaux (1988) devises a diagnostic for intermediate copies in CP of successive cyclic wh-movement based on the interaction between binding and reconstruction, a diagnosis that Fox (1998) extends to copies adjoined to the verb phrase. Consider (17). The relevant potential reconstruction sites are indicated by underlined asterisks/checkmarks (data from Fox 1998: 157). (17) Reconstruction to vP √ a. [Which of the papers that hei gave Maryj ] did every studenti ask herj to read ? carefully? b. *[Which of the papers that hei gave Maryj ] did shej ? ask every studenti to revise ? ? These examples are interesting in that the wh-phrase contains both a pronoun, “he”, which must be bound by “every student”, and an R-expression, “Mary”, which must not be c-commanded by the coreferent pronoun “her/she”. Thus, the wh-phrase must reconstruct to a position below “every student” and above “her/she”. In (17a), such a position is available, if we assume that the whphrase left an intermediate copy adjoined to the verb phrase [ask her to read], and indeed, the sentence is grammatical (if a bit awkward). In contrast, (17b) has no such position is available. In order for “he” to be bound by “every student”, the wh-phrase must reconstruct to its merged position, and yet in this position “she” c-commands “Mary”, violating binding Condition C. Thus, the sentence is ungrammatical. Unfortunately, this test could not be applied to unaccusative verbs, due to the number of necessary internal arguments. However, we can apply it to passives. In (18a) and (18b), the relevant situation is one in which Mary keeps being introduced to her own date at parties. In (18c) and (18d), one should think of a charity auction at which dates with bachelors are sold. (18) Reconstruction to Passive VP a. [At √ which of the parties that hei invited Maryj to] was every mani introduced to herj ? ? b. *[At which of the parties that hei invited Maryj to] was shej ? introduced to every mani ? ? c. [At √ which charity event that hei brought Maryj to] was every mani sold to herj ? ? d. *[At which charity event that shej brought Johni to] was hei ? sold to every womanj ? ? Identically to (17), the sentences in (18) contain a wh-phrase which must reconstruct below “every man/woman” in order for “he/she” to be bound, and 9 above “Mary/John” for the construction to obey binding Condition C. Again, in the (a) sentences such a position exists, if one assumes that the wh-phrase left a copy adjoined to the verb phrase.9 The fact that (18a) and (18c) are grammatical (although awkward), thus strongly supports the claim that successive cyclic wh-movement proceeds through passive verb phrases, as well as unergative ones. In (18b) and (18d), no reconstruction site exists that will satisfy both binding conditions at once, and the sentences are ungrammatical, as predicted. We thus have another potential argument for the phase-hood of passive verb phrases. Interestingly, when we turn to parasitic gap constructions, we fail to observe the now-familiar pattern of passive and unaccusative verb phrases patterning with unergative vPs. Nissenbaum (1998) argues for an analysis of parasitic gaps whereby a vP-level wh-trace is crucial for the interpretation of these constructions. (19) illustrates the normal composition between a vP-adjoined adjunct and its host vP,10 using Predicate Modification to create a conjoined interpretation (see Heim & Kratzer 1998). (19) The Interpretation of vP-adjoined Aduncts John filed the paper [PRO without reading it] <t> TT T H HH (John) filed the paper <t> HH H TT <t> T without PRO reading it Given operator movement creating a lambda abstract in an adjunct containing a parasitic gap, however, the adjunct is unable to compose with its host vP: (20) Type Mismatch Created by Operator Movement in PG Adjuncts John filed the paper [OP [PRO without reading tOP ]] 9 This discussion assumes that the “at DP” phrases are merged as the lowest argument in the verb phrase. Pesetsky (1995) argues for such a “cascade” structure for these adverbs based on binding and NPI licensing. Since it is binding that we are concerned with here, the assumption seems relatively unproblematic, particularly since the contrast in judgements between the (a) and (b) sentences in these examples would otherwise be difficult to explain. 10 Nissenbaum demonstrates that the tests which support a Kaynian, or “cascade”, structure for many “adjuncts” argue for a right-adjoined, or “layered” structure, for the adjuncts found in parasitic gap constructions. 10 <t> TT T <?? > HHH (John) filed the paper HH HH < e, t > TT T OP without PRO reading tOP Nissenbaum’s key insight is that the structure would be interpretable if: (i) a whphrase from the main vP moved to adjoin to vP, creating a lambda abstract;11 and (ii) the adjunct clause containing the parasitic gap merged counter-cyclicly just below the root. (See Nissenbaum for a demonstration that this analysis accounts for the familiar properties of parasitic gap constructions, as well as previously unaccounted for or unnoticed properties.) (21) Type Mismatch Resolved by Movement to vP and Counter-cyclic Merger Which paper did John file [OP [PRO without reading tOP ]] HHH VP < t > which paper HH HH H HHH VP < e, t > VP < e, t > HHH λ1 <t> TT T HH HH H TT < e, t > T OP without PRO reading tOP (John) filed twh Therefore, the parasitic gap construction can serve as a diagnostic for a wh-trace adjoined at the level of the verb phrase. Applying this diagnostic to unaccusative verbs again proves impossible due to the number of internal arguments needed. Applying the test to passives requires use of an overt subject in the subordinate clause, since PRO in these 11 Actually, short distance A-movement to vP also creates the relevant lambda abstract, thus the licensing of parasitic gap constructions by short distance scrambling in German, Dutch, and Hindi. 11 adjuncts, with or without a parasitic gap, seems to strongly resist being controlled by a passive subject, instead prefering to be coindexed with an external argument of the host verb phrase. This change makes the constructions with unergative verb phrases slightly marginal, however the corresponding examples with passive verb phrases are ungrammatical.12 (22) Parasitic Gaps are Ungrammatical with Passive Verb Phrases a. ?Which house did John buy [OP [before we had a chance to clean tOP ]]? b. *Which house was John sold [OP [before we had a chance to clean tOP ]]? c. ?Which house did John buy [OP [without our having fixed up tOP ]]? d. *Which house was John sold [OP [without our having fixed up tOP ]]? e. ?Which book did John buy [OP [in order for his wife to have a look at tOP ]]? f. *Which book was John sold [OP [in order for his wife to have a look at tOP ]]? Here we find a genuine difference between passive and unergative verb phrases. If this difference is attributed to a lack of a wh-trace at the level of the passive verb phrase, the data considered in (18) above becomes quite mysterious. The only alternative explanation I can suggest, however, is that the difference be attributed to some yet ill-understood thematic requirements of the adjunct. Clearly more research needs to be done on this topic. 4 Conclusion In this squib, I have considered the possible status of the unaccusative and the passive verb phrase as a phase. I have demonstrated that, parasitic gap constructions aside, all applicable tests treat passives and unaccusatives on a par with unergatives. My conclusion is thus that if unergative verb phrases constitute a phase, then passive and unaccusative verb phrases do as well. One of Chomsky’s motivations for claiming that passives and unaccusatives form phases is his desire for selection of a lexical sub-array to be computationally trivial. The intuition is that if a phase can be defined by a finite number of functional categories13 , than the selection of the lexical sub-array will be greatly simplified. One way to reconcile the evidence considered here with this goal is 12 I owe great thanks to Jon Nissenbaum for clarifying my understanding of parasitic gaps and for assistance with the examples. 13 Chomsky left open the possible phase-hood of DPs and PPs. 12 to argue that passive and unaccusative verb phrases are also dominated by a vP. Indeed, Baker, Johnson & Roberts have argued that passives contain an external argument, which they identify with the passive morpheme. Independently of their particular implementation, it is plausible that passives have an external argument, which in the Minimalist theory assumed would require presence of v to introduce this argument. Furthermore, Chomsky (1998a,b) has suggested that v may be needed to account for the word order facts of the unaccusative verb “seem”. (23) Possible Evidence for v with Unaccusative Verbs John seems to me to be happier these days. Thus, in (23) we see that “seems” has moved past its dative subject “to me”, plausibly to the head of vP. The claim that v dominates all verb phrases of course requires a distinction be made between two flavours of v, one that introduces an external argument and one that does not. This loses the original appeal of the v proposal which, by attributing to v the introduction of an external argument and checking of accusative case, allowed a neat statement (if not a true explanation) of Burzio’s Generalization. Whether or not this appeal is empirically justified must be left to future research. References Baker, Mark, Kyle Johnson, & Ian Roberts. 1989. Passive Arguments Raised. Linguistic Inquiry 20:2. 219-251. Bouton, Lawrence F. 1970. Antecedent-Contained Pro-Forms. In Proceedings of the Sixth Annual Meeting of the Chicago Linguistics Society. Bresnan, Joan. 1972. On Sentence Stress and Syntactic Transformations. In Michael Brame (ed), Contributions to Generative Phonology, 73-107. Austin: University of Texas Press. Chomsky, Noam. 1970. Remarks on Nominalization. In Roderick A. Jacobs and Peter S. Rosenbaum (eds), Readings in English Transformational Grammar. New York: Ginn. Chomsky, Noam. 1998a. Class lectures. Chomsky, Noam. 1998b. Minimalist Inquiries: The Framework. Distributed by MITWPL. Chomsky, Noam & Howard Lasnik. 1993. Principles and Parameters Theory. In J. Jacobs, A. von Stechow, W. Sternfeld, & T. Vennemann (eds), Syntax: An International Handbook of Contemporary Research. Berlin: Walter de Gruyter. Fox, Danny. 1998. Economy and Semantic Interpretation. MIT dissertation. 13 Heim, Irene & Angelika Kratzer. 1998. Semantics and Generative Grammar. Cambridge, Mass.: Blackwell. Lebeaux, David. 1988. Language Acquisition and the Form of the Grammar. UMass dissertation. Nissenbaum, Jon. 1998. Movement and Derived Predicates: Evidence from Parasitic Gaps. In Uli Sauerland & Orin Percus (eds), MITWPL #25 The Interpretive Tract. Cambridge, Mass.: MITWPL. Pesetsky, David. 1995. Zero Syntax. Cambridge, Mass.: MIT Press. Pesetsky, David. 1998. Class lectures. Urribe-Etxebarria, Myriam. 1995. Levels of Representation and Negative Polarity Item Licensing. Proceedins of the Fourteenth West Coast Conference on Formal Linguistics. 571-586. 14
© Copyright 2026 Paperzz