Download as pdf
Download as pdf
You are on page 1of 10
an. Introduction to Government and Binding Theory “The English equivalents of {80) are far less acceptable: xr brother, to whom I wonder which stories they told, task which you ignore to whom they'd entrust new idea, of which I can imagine what you think, ‘The reader will probably be able to identify as violations ofthe subjacency condition. In ( been exteacted out of an embedded question introduced by which stores, crossing IP, and IPs 3 English examples 82) your brother [er to whom fou] wonder [which stories they told, ¢ + Si) “A Apparenly his ype of extaction i allowed in Kalan omiting ilevane dail the Sacre of (80a) is (3). A xs has ah, ae a so fratello [op ia mi domando ley he storie fn abbiano raccontato sun) One possibility would be to claim that the subj _guage-specific, ike the doubly filled COMP iter, and does not spply in Talian If chs were true one would equally expect that extraction out of any type of inditecr question and +0 fact (example from Rizzi, 1982 4 Yeo fratello, [er a eu Lo Ja possibiltt [ert your brother whom fear the possiblity che fp abbiano raccontato % tuto) that they-have told everything [Rizzi's proposal to account for the example given here is NOT that basi of this proposal the grammaticality of ( WHemovement 413 examples only one GP has been crossed (se (83)). The ungrammaticality of (84) is also predicted, In (85) we indicat the celevant bounding nodes by the dacriic # 85 "tuo fratello, [op & ct fy temo Ine la possbiit[c € che [wabbiato # ¢ raccontaro ¢ tutto) In (85) che PP a cui (to whom’) originates in the lowest clause: ic the complement of racontae (ell). cis fst moved tothe low , fand then it has to move accoss CP and across NP, costing 9¥0 nodes “Toe gbjaeny maa is one of he cars formulate nd poem theory. 7 Binding Theory and Traces of Wh-Movement 7.1 Typology of NPs In the discussion of roh-movement it bas become clea that some of the wh- ‘aces (B6) have the status of NPs: 862 Whom, will Lord Emsworth invite «? Mid) Which devectives; do you expect [t; to admice themselves, most)? normally taken by an NP, ies case: woased by seta cole. Tn (860) the race binds a relexive with which t shares features of person, oumber and gender. If thse wh-taces are NPs" the next question is how they behave with respect az those traces? Tn rs [ Anaphor) and ae. where itis assigned NOMINATIVE case, (In passing we draw the reader’s attention tothe fact thae i x quite posible, asin (91), that a wh-constituent ject porition robo is then wh-moved to ignore the cae diagnostic and assume to our detriment. An ND-zacs is subjex to Principle A of che BT: must be boun in its GC. The GC is the lower | finite clase (ce chapter 4 fr the definition of the GC and for the role of | AGR in particular) and clearly the wace is there. Whatever sy | tactic repeesentation we imagine for (908) it wil vilate some principle of 5 The en is fom May (1985), four grammar. (90b) has no legitimate syncactic representation and is ‘ungeammatical. 7.2 Crossover rure™ the following examples have received a lot of attention cf 93a "Who, does be; think left? 93b "Who, does he, think you saw 1? 986 Who, does be set? ‘The umgrammaticaity was at one time attributed to the fact that wh ‘movement moves a constituent across a coindexed pronoun Ir was proposed 94a Lefeness condition we ‘A.wb-tace cannot be coindexed with a pronoun to its et. 94 "Who, does hey think t let? "Now consider the contrast in (95). An example such as (956) is referred to as weak crossover chosen because ede Uh syoce ‘0 ve senienes, for instance, moves an object NP into = ig moras scx NP of» nosnte caoetoe higher clus. In both cases we aomume hat the esbject ed by he moved of ence-fnal po- sition which can host the moved NP, whether this position be categorally specified as an NP position, or caegorially unspecified. We conclude that the [NP is moved to a position created fort. In other words the moved NP is ‘adjoined. (See the discussion of adjunction in connection with structure ‘preservation in chapter 6; and also section 3.4 above.) Let us assume that the moved NP is adjoined to VP producing a structure like in (100): 100 vP ‘The moved NP, commands its trace, 1 desirable result since we have seen Obviously, the adjunction analysis is incompatible with fe hypothesis that the chain 7; YP adjoined hss Jed.us to redefine the notion dominance and also to intodce the concept exclusion: - 6 Dominance ‘Ais dominated by B only if A is dominated by every segment of B. 7 Exclusion B excindes A if no segment of B dominates A. 10 Exercises Exercise 1 Discuss the motivation for the intermediate traces in the following representations: a Who, does the detective think [rt that f» he likes t best? 1b Who,does the detective thnk fe tile 4 tikes him best? Exercise 2 Consider the folowing sentences, some of which we have already iven in the exercises to the Introduction of this Eook. 1 Pich mando you word when Spay wt yp 2 Tio wo yu woot wh presen avete 3 72Wihich present do you wonder wi wi 44 72Which man do you wonder whe 5 Whicn man do you wonder wheter 5 Wich man do you wonder wall giv to John? 7. ‘Which man do you wonder wher invite John? ‘Attnough none of the sentences above is entirely acceptable to all native speakers, the degree of unacceptably varies. A sentence with an asiarisk is worse than one wth a question mavk. Try 10 ‘account for the relative unacceptably of these sentences using the ‘concepts developed in chapter 7. When discussing these sentences: you should first of all determine their syntactic representations, D-siroture and S-structure. Then you should tr to identity which principle or principles are violated. In your analyels you will no doubt discover thatthe extraction of a ‘subject whconstituent Is consistonty more dict than that of an object. This type of asymmetry was discussed interms of the that trace fiter. From te analysis of the examples shove try to check —wather efter as formulatod in the chapter is adequate and if not, ty i In subsequent chapters we stall rotum to exam plos such as those above. If you are a native speaker of a language other than English then check how translations of examples tke those above fare in this language. Exercise 3 In section 6.3.1 we have seen that an analysis cf lft dislocation in terms of mavement and pronoun insertion is not consistent with our present version of the grammar. A structure that losaly resembles loft cislocation is topicalization: na5 a8 Introduction to Government and Binding Theory 1. **Which book did you wonder when would be published? 2 "Who did Poirot tell you why he had interviewed? How could we account forthe fact that ee than (2)? Consider the following data trom Italian (Rizzi, 1962b: 54-6). Can the subjacency parameter discussed in section 6.4 account for hem? 3 Non 80 proprio chi possa avere indovinato a chi ‘not. know really who could have quessed to whom affider® questo incarico Vivenitust this task ‘L realy don't know who might have guessed to whom | wil ot this task” 4 “Questo incarica, che nonso proprio chi possa avere this task, "which not know realy who could have indovinato a chi affderd, mi sta creando un sacco guessed to whom entrust, me is giving alot i gratiacapi. ‘which believe that you know to whom ho dedicato, mi sempre stato molto caro. have dadicated, to-me is always been very precious ‘My first book, which | know that you know to whom I have dedi cated, has always been very dear to me.’ 6 “limi primo libro, che soa chi credi che abbia ‘my first book, which How to whom you-tink that -have dodicato mi @ stato sempre molto. caro, dedicated,to-me is been always very dear. Jaoggli (1981: 170) gives the folowing Spanish data: 7 “etunico encargo que no sablas a quid Iban the only task which you didn'vknow to whom they-would 2 dar give @ “Aquign no sabias que le regalaron? to whom didn't you-know what thay-had-alven ‘Wiomaverent aw © “tu hermano, a quién me pregunio que. historis fe habran your brother, to whom — l-wonder what stories they have contado told, ‘Assuming that the bounding nodes for subjacency may be paramettized (along the lines suggested by Rizzi 1982b), would the Gata above suggest that the bounding nodes in Spanish are lke in English or tke in Nalin? Exercise 9 In section 5.1 we discussed the vacuous movement hypothosis for ‘subject extraction. Consider the following examples from Chomsky (19860: 50, example (109). According to Chomsiy (1) is more ao- coplable than (2). Would this contrast in grammateaity be relevant to the discussion In section 6:17 11 He is the man te whom | wonder who knew wiich book to give. 2 We is the man to whom I wonder who John tld which book to give. In chapter 5 (section 9:2) wo discussed the positon of whether as ‘CP. According to Chomsky (19860: 60), ‘more acceptable than (2) above. 3 He is the rman to whom I wonder whather Job told us which book to giva. 'Doos this throw any light on the discussion in chaper 5? 2 Insroduction to Government and Binding Theory 34.3 Constraints on the interpretation of the non-overt object 3.44 The identification of the empry category 4 Nomovert antecedents of wh-movemeat 4.1 Relative clauses 4.1.1 Empty operators and object relatives 41.2 Subject relatives adjuncts C and operator binding 4.3 Null objects in Portuguese Parasitic gaps 5.1 Description 5.2 The PRO hypothesis 5.3 Parasitic gaps are traces 5.4 Conclusion 6 Summary 7 Exercises Introduction and Overview (Chapters 5-7 inteaduced three types of noa-overt NPs, !RO, NP-trace and twh-trace, and at this point it is useful ro cake stock of the various properties that we have attbured to these non-overt or null elerats. “The fist part ofthis chapter provides a survey of our diswsion and extends ‘our analysis ofthe properties of aul elements. We introduce a principle 10 ‘constrain the occurrence of traces: the empty category rcnciple. In the second part we complete our inventory of null ehments. We discuss evidence for a fourth null element, the zero pronoun pro, We also discus the null operator and dlostrate ics use in English and in Potuguese. ‘Since much ofthis chapter isa revision sion ofthe preceding pact fof the book, the reader should find this which enables him to consolidate his kx ‘subsequent part of the book in which we chall farther develop the grammar wich we have been elaborating. ‘Section 1 contains an inventory of the empty categores discussed 30 far. Section 2 discusses the licensing of empty categories and the empty category principle, In section 3 we deal with the null element pro. Section 4 discusses ‘non-overt antecedents of r-traces and section S deals wib the parasitic gap ‘phenomena. 1 Null Elements in English: Recapitulation “The null elements introduced so far are exemplified in (2). (1a) illustrates PRO, (tb) ilustrates NP-race, and (te) illustrates wh-race. Ja John, would prefer very much [[PRO, to invie Bill]. bf Bil will be invited «J. ; Le [ap Whom) would [Joka pefer [yt Hor [p us to invite ei? “The thee null elements iiustrateé here all have an S-structure antecedent, an [NP with which they are coindexed, but the status of PRO is quite distinct fiom the status of traces, And similarly, NP-trace and wh-trace can be identity another null element which is governed: pro. 1.4 The Binding Theory and the Typology of NPs ‘We have established thatthe three null elements posited are specified forthe features [tAnaphor] and [¢Pronominal] in the following way: 4. Typology of NPs Type NON-OVERT {Anaphor, —Pronominal NParace [Anaphor, +Pronominal] 2 [-Anaghor, ~Pronominal) here [+Azaphor, +Pronominal PRO PRO is subject to both Principles A and B, hence its special discussion in chapter §). NP-trace and PRO both have the cing [+Anaphor], in contrast with wh-tracss. 1.5 NP-trace and PRO 1.5.1 CHAIN FORMATION In chapter 6, section 4.6.1 we discussed the chain formation process, We recapitulate the discussion briefly here. Consider (Sa): am eevenssry vp ingty Ssngsnns 431 Se Poizo, seems Ly § to enjoy the enquiry). Consider now (6a) in which PRO is the subject of the non-fnite clause: 64 Poirot, wants [er [y PRO| to write @ novell) Both wat and write asign an external theta roe. As befoe se ignore the subjeevin-VP hypothesis which dose not slr the line of agumen (ee foot note 2}. By the chin formation process, we cat form ete he chain (6), ‘vhere PRO and Poros each hed chain, o (6, where Poot and PRO are part of singe chin: 6 Poirot, PROP 6 Poirot, PRO> ‘The thet cso es [G3 be cain Port, PRO cnsine wo ‘eae, Foret “Fairs, and PRO, each have 10 belong to « Me ~/ Needless to sy, all these properties are 1.6 Traces [NP-trace and wh-trace share the feature {-Peonominal] (ef. section 1.4). Daring the discussion in chapters 6 and 7 and the sections above we have already identified several properties which distinguish NP-araces and wh-traces. These concern the moved clement, is landing site and its extraction sie: 9) Traces: survey of propesties NPara00 wie Moved category: Ne XP (NP, PP, ete.) Landing site ‘Acpostion by subetiuton “A‘postion by substuton ‘or adjunction NP-postion Spee, CP} or adoined postion Properties of antecedent case Yes No Chain mohar Kechala ‘Proportis of trace Features FeAnaphor Anaphon Prenominal Pronemina] Binding theory a 8 Tota ola Yes You case No ‘Ys (aon target = NP) Governed Yes Yes invite the reader to try and relate them in Aiscussion of NP-teaces in chapter 6, 2 Null Elements in a Grammar 1f-we assume. mull cements. a5. component of the. grammar_of natural languages, we must assume that the language learner has talate such null elements in the representations he assigns to sentences, He needs to have evidence for positing these categories and ways of identifying ed from chapter 3 that overt NPs must be asiged abstract case, We Jeatute on the NP makes it visible for theta role say that abstract case legitimstes NBs, of, to put it i=) formally licensed by case, Null elements too\% ) in ungovemed contexts. We can say that ‘contexts; nd (i) that its interpretation is detexmined by control theory. In the next section we addeess the licensing conditions of traces. 2.1 Formal Licensing: the Empty Category Irinciple In our survey ofthe properties of traces we have and wh-waces are governed. Government has licensing condition for traces: races must be govemed. However, simple goverment will not suffice to license a wace: traces must be governed ina special way. "The discussion of the government requirement of races starts out from the subject-object asymmetry exhibited in (10) and (11) andalready discussed in chapter 7, ection 5.1 10a Whom, do fp you think [cyt that [p Lord Emsworth wil invite «]}? “Whom, do [p You think [er Lord Emsworth will invitee}? Who, do fy you think fort that [pt will invite Poirot]? “Who, do [p you think fer [a 6 will invite Poiro]? “While objects can be freely extracted across ovex complementiers, sub- jects can only be extracted from clauses without overt complementizers. This ofthe thatace feet. The thattace Biter erpreted in the light of the government requirement for traces is that in order to be licensed, i.c, legitimated in certain positions, In (152) and in (15b) the lowest trace is governed by the verb invite by which it is also theta marked. This type of government is theta-government. We shall be looking ot intermediate traces in chapter 9. ‘of the theory of nall elements, The ECP isa principle which, at the moment, does not seem to follow from anything else inthe grammat. Hving established that the ECP formally licenses traces we now turn to the question how their content is recovered. There isa rather natural answer . ‘en that traces must have an antecedent. Tae properties of the 3 able to be recovered by virtue of the coindexation with its Ta the next secsions we illustrate the ECP and we turn to some problems 2.2 Subjacency and ECP ‘wonder when Jobn boughs? Subj teincly vols "Who wonder when bought these books? “1, 182 ?What do you wonder who will read? 186 *Who do you wonder what will read? Although none ofthe sentences in (1 is a marked decrease in the aoceptal the S-structure representations of these se intuitions 190 tay What do fy, you wonder [on when; [apJobn bought x)? ? ‘ 196 “Lon Who, do {oy you wonder fo, when; tna bought thse books eI? ’ ‘ 20a 2m What, do fp; you wonder [cm who, fr § will read tJ)))? ‘ s 205 *leer Who, do [ox you wonder [ce what, ox § will ead tI? # 4 Allthe examples above are violations of Ros” slrisland constrain. In more general terms they at subjacency: two bow subjacency condition ig violated in a similar fashion, but the sentence is “es of mame anc rah gd iy thew, Conse 232 What John will do is {yp fix the car clumsily). 23b_ [yp Fox the ear clumsily) is what they all do. 23e [yy Fox the car clumsily] John surely did, ‘n.the thse examples above the manner adjuncts behave as VP constiu- «us. In the pseudo-cleft sentences (232) and (236) che adjunct coms sclefied ‘wit the VP. In (23c) the mannes adjanetis preposed along withthe verb and ts object. Under this analysis the Sstrocmre representation of (22) would be “24: cay o Spee Cl —, a! ° 1.20 ithe nt sepsentaion eal then the stew hold i oe “Although adjunc race and subjece races are simular in chat they ean only satisfy dhe ECP through antecedent goverment, the wo umes of traces pat~ tern differently with respect to long movement. Compare long subject swh- ‘movement in (25) with long adjunct wh-movement in (26): 52) Wien do 08 ik a wot wiles When, do fp you think [— «Tp Emsworth wil invite Poior eI? 7 ch probe, na John could sole this wy} 1 wonder (on which probeza, C? [ma Jobs could ‘moves tothe matrix [Spec, CP,} and which problem movesto the lower (Spec,

You might also like