Quantifier Float and Wh-Movement in an Irish English James Mccloskey

Quantifier Float and Wh-Movement in an Irish English James McCloskey

TheEnglish of northwestern Ireland allows quantifier float of aprevi- ouslyundocumented kind in wh-questions.The quantifier all, though construedwith a fronted wh-pronoun,may appear in apositionconsid- erablyto the right of that pronoun. It is argued that all so stranded marksa positionthrough which a wh-phrasehas passed or in which a wh-phraseoriginates. The construction then provides visible evi- dencefor intermediate derivational stages. This evidence is used to developa newargument for successive cyclicity and to argue for overt objectshift in English and for an originsite for subjects strictly within VPandbelow the object shift position.

Keywords: quantifierfloat, wh-movement,successive cyclicity, object shift,verb raising, VP-internal subjects

1ALocalEnglish Thisarticle examines part of thesyntactic system of alocalvariety of English —avarietyspoken inan areawest and east of theriver Foyle in the northwest of Ireland. The area includes at least Derry city,the Inishowen peninsula, southeast Donegal, and the westernmost parts of counties Tyroneand Derry. At therisk of sacrificingaccuracy for brevity,I willcall the English(es) spoken in this area WestUlster English. WestUlster English is close to the east Ulster variety (called BelfastEnglish )describedin theimportant body of workon dialectsyntax developed by Alison Henry.The twovarieties are, however, distinct in numerousways —phonological,morphological, andsyntactic. In particular, the phenomenon considered here does not occur, as far asI know, inBelfast English. 1

Iwouldlike to dedicate thisarticle tothe memory ofmy teacher Lee Baker,who, among many other things, was thefirst, I believe,to dotheoreticalwork on the syntax of anIrish variety of English(Baker 1968:66). Thanksto Sandy Chung, Norbert Hornstein, and Jason Merchant for comments onearlier versions.The research reportedon here grew outof conversations with Cathal Doherty. An early versionwas presentedat NELS26 (Harvard andMIT). Versions were alsopresented to audiences at theUniversity of California, Santa Cruz, the University of California,Berkeley, and Cornell University. I am especially gratefulto three reviewers for LI forproviding unusually thoughtfuland useful critiques of an earlier version.The article was preparedwith the help of research fundsfrom the Academic Senate oftheUniversity of California, Santa Cruz. 1 West Ulster Englishis, or was, mynative dialect. The observations presented here, however, derive from work with15 speakers inall (directlyor byproxy).I am particularlygrateful to Brian McCloskey, Martin McCloskey, Elizabeth McCloskey,Cathal Doherty, Frank McGuinness, Jonathan Allison, John Dunnion, Elaine Brotherton, Mary McLaughlin, BillyRobinson, Da ´ith´õ Sproule,Ciaran Tourish,and Paul McGill for their help. Thanks also to Karen Corriganand to AlisonHenry for their help in delineating the geographical range of thefeature.

Linguistic Inquiry, Volume 31, Number 1,Winter 2000 57–84 q 2000 bythe Massachusetts Institute ofTechnology 57 58 JAMESMCCLOSKEY

Workingon localvarieties of Englishis adifficultbusiness (see, e.g., Henry 1995:12 –15). Thestigmatization that nonstandard varieties are subjected to makes it even more difficult than itusually is to establish reliable data. Luckily, though, the syntactic features I willdiscuss here havenot, as far asI know,been noticed before— by linguists, by speakers, or bythe guardians ofpurity. Since the features have escaped notice, they have also escaped condemnation. Many speakers,in fact, express surprise when it issuggested to them that the features in question are nota partof ‘ ‘Standard’’ English.This fact has the happy consequence for thelinguist that at leastsome of thedifficulties that usually make working on nonstandard varieties so fraught do notapply in the present case.

2ThePhenomenon Thebasic observations can be quicklymade. Most varieties of English(although not all, it seems) allowquestions of thekind shown in (1). (1) a.What all did you get t for Christmas? b.Who all did you meet t whenyou were inDerry? c.Where all did they go t for theirholidays? Suchquestions differ from thosein (2) inimplicating that the answer is a pluralityand in insisting onan exhaustive, rather than a partial,listing of themembers of theanswer set. (2) a.What did you get t for Christmas? b.Who did you meet t whenyou were inDerry? c.Where did they go t for theirholidays? Inaddition to (1), though, West Ulster English allows (3). (3) a.What did you get all for Christmas? b.Who did you meet all when you were inDerry? c.Where did they go all for theirholidays? Thequantifier all in(3a– c) isconstruedwith the interrogative pronoun and not with the subject. Thatis, the examples in (3) aresynonymous (completely so, as far asI havebeen able to tell) withthose in (1). 2 Therelationship between the wh-pronounand its associated quantifier is indicatedin (3) bymeans of underlining, a notationaldevice that I willadopt throughout. The effectoccurs in both matrix and embedded questions. (4) a.I don’t rememberwhat all I said. b.I don’t rememberwhat I saidall. Iwilloccasionally refer tothe construction exemplified in (3) and(4b) as wh-quantifierfloat.

2 Why all and how all are bothimpossible. I suspectthat this must ultimately reflect thespecial denotationalproperties of why and how. Szabolcsiand Zwarts (1993)argue that how is anelement thatranges over domains whose elements exhibita partialordering; who,what, where, and when rangeover individual domains. For why, see theirfootnote 14. When all seems tobe marginally possible. Q - F L O A T A N D W H -MOVEMENTINANIRISHENGLISH 59

Thepurpose of thepresentarticle is to usethis construction as a probeto investigateproperties of (long) wh-movementand to explore some aspects of clausal organization.

3 Basics Wh-quantifierfloat bears a clearfamily resemblance to quantifierfloat of themore familiar sort, illustratedin (5). (5) Thechildren must all have gone to bed. Themajor difference between (5) and(4b) is that in theformer theDPconstruedwith all occupies anA-position, whereas in the latter it occupies an A ¯ -position. Twoprincipal ways of understanding the syntax of quantifier float have emerged in recent years.One originates in workby Sportiche (1988) and was developedsubsequently by himand byothers(Giusti 1990, Shlonsky 1991, Sportiche 1996, Merchant 1996). It holds that (5) derives from arepresentationin which thechildren and all form aconstituent.The other takes all to be anadjoined adverbial element that has the special property that the constituent it adjoins to mustcontain a trace(Klein 1976, Williams 1980, Kayne 1984:chap. 4, Dowtyand Brodie 1984, Miyagawa1989, Doetjes 1992, Baltin 1995, Torrego 1996, Bobaljik 1995, 1998, Morzycki 1998). Iwilldevelop an approachof thefirst type here— atleast for wh-quantifierfloat —justifying itas the discussionproceeds. Whether this approach is also right for themore familiar A-movement casesis a differentquestion, which I willconsider (but only very briefly) in section 7. Saywe beginwith the assumption that who,what, where, and when arepronouns and that theinternal structure of, for instance, who all isanalogousto that of they all. FollowingPostal (1974:111)and Koopman (1999), we mightalso assume that as (6) illustrates,the order [pronoun quantifier]derives from astructurein which the quantifier precedes the pronoun.

(6) a. DP b. DP

DPj DPj

D DP D DP they who

all tj all tj Given(6), phrases like who all havean internalstructure in which a certainambiguity of factoriza- tionwill hold in anypotential application of wh-movement.The lower DP ( who)of(6b)evidently bears a wh-feature.It should therefore be able to undergo wh-movement.But from thegrammati- calityof (1), we knowthat movement of the entire phrase who all alsoresults in successful checkingof the wh-feature.Let us for presentpurposes take the traditional view that this is possiblebecause the wh-featureis instantiated both on the specifier who andon the dominating DP (who all), as in (7). 60 JAMESMCCLOSKEY

(7) DP [wh]

DPj [wh]

all tj

who If otherconditions are met, then, both (1) and(3) shouldbe possible. Prominent among these ‘‘otherconditions’ ’ arethe locality requirements on movement. In the theory of localityof move- mentdeveloped in Chomsky 1995:311, 1998:38, a targetK maynot attract an element b if there isanelement a closerto Kthan b ,whichcould enter into a legitimatefeature-checking relation ifraisedto K.‘‘Closeness’’ isdefinedin termsof asymmetricc-command: a iscloserto K than b ifK c-commands a , a c-commands b , and b doesnot c-command a .Bythisdefinition, neither oneof thetwo wh-DPs of(7) iscloser to thetarget of wh-movementthan the other, since neither c-commandsthe other. Both should, then, be accessibleto C. 3 Viewedin this general light, (3) isa grammaticalviolation of Ross’ s (1967)Left Branch Constraint.The question arises why (3) isnotpossible in allvarieties of English.I willreturn to thisissue. Note, though, that the phenomenon now falls within the known range of syntactic variation,since we knowthat some languages permit extraction of a wh-specifier,stranding the head,and that some do not (Ross 1967, Bresnan 1976, Uriagereka 1988, Chomsky 1995:263, Aissen1996, Chung 1998:255 –257,308 –313,Kennedy and Merchant 1998). 4 Finally,I shouldpoint out that examples such as (3) havecertain very distinctive prosodic characteristics.The sequences get all in (3a), meet all in (3b), and go all in(3c) areprosodic unitswhose most prominent element is theverb. There is a strongintonational break following thisprosodic unit. These facts are discussed in some detail in McCloskey 1998, and an analysis isproposedthere. For presentpurposes, the core observation is that wh-quantifierfloat examples

3 Theidea that,when wh-pronounsalone raise, theymust pass throughthe specifier of all isin harmonywith the bodyof evidence that some generalprinciple forces movementout of DPtoproceedthrough the specifier positionof D (Cinque1980, Torrego 1986, Stowell 1989, Giorgi and Longobardi 1991, Valois 1991, Szabolcsi 1994, Aissen 1996). Thelogic of thetext discussion does not force thisconclusion since itwouldapply equally well toany wh-DPembedded withinanother. Obligatory extraction through the specifier ofDPwouldbe ensuredif DPs were ‘‘phases’’ inthe sense ofChomsky (1998). This in turn would be guaranteed if phases correspondedto constituents that were saturated,or ‘‘closed’’ inthe Fregean sense. 4 German has aconstructionthat is at least superficiallysimilar tothe West Ulster Englishconstruction considered here. (i)Was hastdu alles gekauft? whathave you all bought ‘Whatall didyou buy?’ TheGerman constructionhas beenstudied by Giusti(1991), Beck (1996),Pafel (1996),and Almy (1997) and especially inthe very careful anddetailed study by Reis (1992).Pesetsky (1998:48) discusses aninterpretive parallel between the German andWest Ulster Englishconstructions. Q - F L O A T A N D W H -MOVEMENTINANIRISHENGLISH 61 areoptimal when all canbe incorporatedinto a precedinghead, preferably a verb.The requirement isnot absolute (as we willsee), but examples that depart from thisoptimal pattern are liable to bejudgedvariably —evokingdifferent reactions from differentconsultants, and differentreactions from thesame consultant in different sessions. Theseobservations will be bothhelpful and unhelpful in the discussion that follows. They willbe helpfulin the sense that they will provide an accountof someof the subtler distinctions thatwill have to bemade.They will be unhelpfulin thesense that for thepurposes of thisarticle (whichare syntactic purposes), the prosodic factors are a sourceof noise.For anyexample that departsfrom fullwell-formedness, we willhave to worry aboutwhether the blemish has its origins insyntactic factors, in prosodicfactors, or in some interplay between the two. Settling this is not alwaysstraightforward, but arguments can be given, I believe,for mostor all of the cases of centralinterest. The logic of the arguments is familiar and straightforward. In some cases it canbe shown that all prosodic requirements are plausibly met but the example type remains ungrammatical.It seems safe toconclude in suchcases that the deviance has its origins in syntax. Inother cases it can be shownthat keeping the syntax constant, but varying the prosodic factors, makesthe example type either more or lessdeviant. In such cases it seems plausible to attribute thevarying grammaticality to prosodic factors. Inany case, despite the complexity introduced, the factors are real and need to be faced sooneror later.

4Strandingunder ‘‘Long’’ Wh-Movement If mygeneral line of thoughtis on theright track, all in wh-quantifierfloat should always mark eithera positionin which wh-movementoriginates or apositionthrough which a wh-phrase has passed.A particularlytelling confirmation of this prediction is that all maymark the intermediate positionsposited by the theory of successive-cyclic movement. Consider examples like (8). (8) a.What all did he say (that) he wanted t? b.What did he say (that) he wanted all? c.What did he say all (that) he wanted t? Allthree variants are possible. Speakers are virtually unanimous about this, 5 andthere is a prefer- encefor (8c) (withintermediate stranding) over (8b) (with stranding in theorigin site), although bothare clearly grammatical. The possibility of intermediate stranding exists for allthe wh- pronouns. (9) a.Where do you think all they’ ll want to visit t? b.Who did Frank tell you all that they were after t? c.What do they claim all (that) we did t?

5 Twoof 15 speakers rejected theintermediate strandingexemplified by (8c)—one consistently, the other intermit- tently. 62 JAMESMCCLOSKEY

Italso exists for nonfiniteclauses. (10)a. What were youtrying all to say t? b.What did you mean all for meto do t? (11)shows the possibilities that arise for complexstructures involving a nonfiniteclause within afiniteclause. (11)a. What all did he say that he wanted to buy t? b.What did he sayall that he wantedto buy t? c.What did he say that he wantedall to buy t? d.What did he saythat he wanted to buy all? As canbe seen in (11b)and (11c), stranding is possiblein both the intermediate positions. There is,once again, a dispreference(stronger here than for thetwo-clause embeddings) for thecase in which all isstrandedin the lowest IP. Finally,(12) illustrates the case of three finite clauses. As usual,such examples tax speakers’ patience and credulity; but once again it is clear that strandingof all ispossible in the two available intermediate positions ((12b) and (12c)). There isagaina dispreferencefor (12d),in which all appearsin the A-position in which the movement originates. (12)a. What all do you think (that) he’ ll say (that) we shouldbuy t? b.What do you think all (that) he’ ll say(that) we shouldbuy t? c.What do you think (that) he’ ll sayall (that) we shouldbuy t? d.What do you think (that) he’ ll say (that) we shouldbuy all? Theseobservations suggest the analysis schematized in (13), in which all hasbeen stranded in anintermediate specifier of CPposition. 6

(13) [C P whatj [IP . . . say [C P [D P i tj all tj ] that [I P he wanted ti ]]]] Givenwhat we haveestablished so far, thisis expected. If theunderstanding outlined earlier is roughlycorrect, then the two options available in the origin site of wh-movementshould also be availableat eachpoint in the derivation at which wh-movementmust (re)apply. 7 That all insuch cases is actually in the specifierof CP (ratherthan adjoined to a VPprojection containingCP, for instance)is suggestedby thedata in (14)and (15). (14) illustrates the case of

6 Withinthe context of Chomsky 1998, it is notobvious whether the intermediate strandingsdocumented so far involvemovement through the specifier ofCPormovement through the outer specifier of vP.Bothinterpretations are availablebecause we willultimately be ledto theview that English has verbraising out of VP, which would place the verb think in(12b), for instance, to the left ofthe specifier of vP.However,the paradigms we willexamine presently (cf.,e.g., the contrast between (14b,d)and (14e)) speak against this interpretation. 7 Itis knownthat extraction of phrases inthe specifier positionof a CPcomplementdoes not give rise toungrammati- cality(Torrego 1986, Chomsky 1986:25 –27,Postal 1997:34– 35). In the system ofChomsky 1986, the result follows fromthe requirement that specifiers ofL-markedcategories are themselves L-marked. (i)the guy that we couldn’t decidehow many pictures of we shouldbuy Therefore,extraction of thespecifier alone(stranding all)shouldbe possible in principle. However, see footnote9. Q - F L O A T A N D W H -MOVEMENTINANIRISHENGLISH 63 a verb (tell)thatselects a DPcomplementand a CPcomplement.Stranding of all ispossibleto theright of DPandto the left of C, butnot otherwise. (14)a. What all did he tellhim (that) he wanted t? b.What did he tell him all (that) he wanted t? c.*What did he tell all him (that) he wanted t? d.?What did he tell his friends /Mickeyall (that) he wanted t? e.*What did he tell all his friends /Mickey(that) he wanted t? Apartiallysimilar array of judgmentsis seenin (15). (15)a. What all did he say to him that he wantedto buy t? b.*What did he sayall to him that he wanted to buy t? c.?What did he say to him all that he wanted to buy t? d.?*What did he sayto his friends all that he wanted to buy t? e.*What did he say all to his friends that he wanted to buy t? Inthis case the selected complements are PP andCP. Thedifferences between (15c) and (15d) (andbetween those and (14b,d)) seem to be basically prosodic in character, having to do with therelative phonological weight of the material that intervenes between V andthe stranded quantifier(see above,and McCloskey 1998). Insum, the stranding possibilities that are found are those that the general line of analysis proposedhere would lead us toexpect. All in wh-quantifierfloat constructions appears in positions for whichthere is considerable independent evidence that they are either positions in which wh-movementoriginates or positions through which wh-movementpasses. 8 We havein these observationsa newkind of argumentfor thesuccessive-cyclic character of long wh-movement.9

8 Anapparently similar paradigmis foundin Standard English with adverbs like exactly and precisely in uses like thosein (i)– (iv). (i)What precisely do youwant? (ii)Precisely whatdo you want? (iii)What exactly do youwant? (iv)Exactly what do you want? This use of precisely/exactly is presumablythe same as thatfound in (v)and (vi). (v)She made exactlyten trips to France last year. (vi)She made tentrips exactly to France last year. Theadverb in thisuse may beseparated fromthe wh-phrasewith which it is construed. (vii)What do you want exactly /precisely? Intermediate placement isalsopossible. (viii)What did he say exactlythat he wanted? Theinterpretive issues here are alittlesubtle, but it seems relativelyclear thatthe triad consisting of (viii)and (ix)– (x) involvesone use of exactly. (ix)What exactly did he say thathe wanted? (x)What did he say thathe wanted exactly? 64 JAMESMCCLOSKEY

5Strandingunder ‘‘Short’’ Wh-Movement Thismuch established, we cantry to understandwhat happens in apparentlysimpler cases —cases ofclause-bounded wh-movementin which all appearsinside VP. (16)and (17) show all strandedapparently in thecanonical object position (i.e., immediately tothe right of V, tothe left of other complements and adjuncts), however the syntax of that positionis bestunderstood. (16)a. What did you give all to the kids? b.What did you put all in the drawer? c.Who did you send all to the shops? d.What do we needall from theshop? e.Who did you meet all up the town? (17)a. Tell me whatyou got all for Christmas. b.Tell me what you’ ve been reading all. Strandingof all inpositions further to the right within the VP isdegraded. This is shown for oneclass of cases in (18)— cases in which the quantifier is stranded to the right of a clausal complement. (18)a. Who did he tell all he was goingto resign? b.*Who did he tell he was goingto resign all? Strandingof all ina positionto the right of adjuncts is also uniformly impossible.

Urban(1999) points out that just contrastswith exactly and precisely intwo ways: itmay notfollow the wh-element, andit may notbe separated fromthe wh-element. (xi)Just what do youwant? (xii)*What just do youwant? (xiii)*What do you want just? Thisperhaps suggests that the separability of exactly and precisely dependson the possibility of (i), (iii), and (vi). If precisely and exactly are headsthat take wh-phrasecomplements andthat permit optional raising of thosecomplements totheir specifier position,then the possibility of stranding in these cases can beunderstood in the same terms suggested above for wh-all phrases.There are difficulties,though. As pointedout by Urban (1999), (xiv) is mysteriouson this account. (xiv)About what do you want to complain exactly? Andas pointedout by areviewer, theungrammaticality of (xv)is alsosuspicious. (xv)?*What did he say yesterdayexactly that we wanted? 9 These observationsleave all themore mysteriousPostal’ s (1974)observation that (i) is impossible.

(i) *[CP Who do [IP theybelieve [ CP to t that [IP thestudents spoke t]]]]? One possibilityis thatwhatever relation(poorly understood) it is between prepositionand governing head that licenses prepositionstranding fails toholdin (i). Q - F L O A T A N D W H -MOVEMENTINANIRISHENGLISH 65

(19)a. What did she buy all in Derry yesterday? b.*What did she buy in Derry yesterdayall? c.*What did she buy in Derry allyesterday? (20)a. What did you do all after school the day? b.*What did you do after school the day all? c.*What did you do after school all the day? Thismuch is fundamentally compatible with what is known about extraction of objects under wh-movement.There is presumablyno doubtthat the ‘ ‘canonical’’ objectposition (immediately postverbal)is apositionfrom which wh-movementis possible. The ungrammaticality of stranding ina positionto the right of adjunctswill follow under either or bothof thefollowing assumptions: (i) Therelevant position is not one through which wh-movementmay pass. (ii)The relevant position is an adjoined position and therefore phrases that occupy it are islands—extractionof a subpartis impossible but extraction of theadjoined phrase itself ispossible. Theungrammaticality of the relevant examples, however, seems to be much stronger than that oftypical adjunct island violations, suggesting perhaps that assumption (ii) is not sufficient as anexplanation. Let us, then, adopt hypothesis (i), thoughwith the understanding that nothing largeis atstake if (ii) also turns out to play a role. 1 0 Moreimportantly, these facts essentially fall into place given the analysis as developed so far. Now considerthe case of verbswith PP complements.Consistent with our general hypothesis, (21)is completelyimpossible. 1 1 (21)a. *Who did you talk all to? b.*What were youlaughing all at? Strandingof all inprepositional object position is slightly degraded when the preposition is adjacentto the verb. 1 2

10 As always, prosodicfactors may alsoplay a rolein the deviance of (19) /(20b–c). Thisis clearly notthe whole story,though. See footnote14. 11 Areviewer pointsout that the ungrammaticality of (21),if interpreted correctly here, implies that extraction out ofPPdoesnot proceed by way ofthespecifier ofPP— aproposaldeveloped originally by Van Riemsdijk(1978) and Baltin(1978). 12 As pointedout by areviewer, there isan interesting contrast between (22)(weakly ungrammatical ifungrammatical at all) andthe facts of combien-extractionin French,which is completelyimpossible from prepositional objects (Kayne 1981a:97). (i)*Combien a-t-il parle ´ avec d’ e´tudiants? how-manyhas-he spoken with of students ‘Howmany students did he talk to?’ Itseems likelythat this contrast is a reflectionof whatever makes prepositionstranding grammatical inEnglish but ungrammatical inFrench. 66 JAMESMCCLOSKEY

(22)a. ?Who did you talk to all (at the party)? b.?Who was helaughing at all? Whenanother complement intervenes between verb and PP complement,the degree of ungram- maticalityby comparison with (22) is more severe. (23)a. ?Who did you give tea to all? b.Who all did you give tea to? c.*Where did you move the books to all? d.Where all did you move the books to? Further,if aprosodicallysubstantial preposition follows the verb, there is also a noticeabledeg- radationas comparedwith (22). (24)a. Who all were yousitting beside? b.?*Who were yousitting beside all? Thecontrasts among these various cases are almost certainly best understood in prosodicterms, inthat the examples that are most consistently acceptable are those in which the material that intervenesbetween the verb and the element all issufficiently insubstantial (in prosodic terms) thatit can be incorporated into the verb along with all (McCloskey1998). We haveseen this effectbefore (see thediscussion of (14)–(15)), and we willsee it again. Itis conceivable that the relevant difference between Standard English and West Ulster English(one disallowing and the other allowing wh-quantifierfloat) is that West Ulster English possessesthe relevant mechanism of prosodicincorporation but that Standard English does not. 1 3

6TheInternal Architecture of VP 6.1Object Positions Thismuch seems reasonably coherent, but a numberof interesting puzzles remain, among them thefollowing. We haveseen that stranding of all isuniformlyimpossible in a positionto the right of VP adjuncts.(25) and (27) contrast starkly with (26) and (28). (25)What did she buy all in Derry yesterday? (26)a. *What did she buy in Derry yesterdayall? b.*What did she buy in Derry allyesterday?

13 Theprosodic organization of West Ulster Englishdoes seem tobe verydifferent from the prosodic organization ofStandardEnglish. However, some larger syntacticdifference between standardvarieties andWest Ulster Englishmight beat workin addition, since itis afeature ofWest Ulster Englishthat prosodically weak functionwords may appear in arangeof right-peripheralpositions that are forbiddenthem inStandard English. (i)He’ s wile richbut. ( 4 ‘Buthe’ s veryrich.’ ) (ii)You’ re wile thingot. ( 4 ‘You’ve gotvery thin.’ ) Q - F L O A T A N D W H -MOVEMENTINANIRISHENGLISH 67

(27)What did you do all after school the day? (28)a. *What did you do after school the day all? b.*What did you do after school all the day? Earlierwe concludedthat examples such as (26) and (28) are excluded because the positions marked by all areneither positions in which the object originated nor positions into which the objectmight have been moved by wh-movement.1 4 But(26) and (28) contrast with examples such as (29), in which all associatedwith an object isstranded in a positionto the right of another complement but to the left of adjuncts. (29)a. ?What did you put in the drawer all(yesterday)? b.?What did you bring to school all (yesterday)? c.?Who did you send to the shops all (last night)? Strandingof all inthisposition yields results that are uniformly worse thanin the case of stranding incanonical object position. The relevant examples are not completely unacceptable, however. Theannotation ? heremeans that some speakers reject the sentences outright, whereas others findthem relatively acceptable (but degraded in comparison with (3)). Thereis also considerable variationin speakers’reactions to (29); the same consultant will often react differently on different occasions.The empirical challenge, then, is delicate. What is required is an analysis that will accountfor thethree-way contrast illustrated in (30). (30)a. What did you put all in the drawer (yesterday)? b.?What did you put in the drawer all(yesterday)? c.*What did you put in the drawer yesterdayall? Theelements of an account are already in place. (30c) is syntactically ill formed. As for (30b), we haveseen that all-strandingis fully grammatical for allspeakers only if all canbe prosodically incorporatedinto a precedinghead —optimallya verb.Structures that depart from thisideal, but meetthe purely syntactic conditions on stranding,are of intermediatestatus— judged variably by differentconsultants, and by thesame consultant at differenttimes. We canattribute the variably degradedstatus of (30b) to this effect, since in this case the prosodic requirement is not met in itsoptimal form. (30c)also violates this requirement (see footnote14). Insum, then, we canaccount for thethree-way contrast by maintaining that (30a) and (30b) aresyntactically well formed, but that (30b) is prosodicallynonoptimal. (30c) is syntacticallyill formedas well as being prosodically nonoptimal. 1 5 Thatmuch granted, the issue to be addressed iswhy (30b) and its like meet all the purely syntactic conditions on all-stranding.If theearlier

14 (26)and (28) also violate the prosodic licensing requirement discussed earlier. Itis unlikely,though, that this constitutesa complete accountof their deviance. It provides no basis forunderstanding the (sharp) contrast between (26)/(28)on theone hand and (29) below on the other. The account to be developed here categorizes (29)as syntactically well formedbut prosodically nonoptimal; it categorizes (26)and (28) as beingsimultaneously syntactically and prosodically ill formed. 15 Itis alsopossible that the degraded status of (30b)reflects thesame dispreference fordistance between wh-phrase and all thatwe saw inthe discussion of long movement above. Compare thediscussion of (8)–(12). 68 JAMESMCCLOSKEY discussionis on track, this must mean that both positions are positions that an object wh-phrase mayoccupy in the course of its derivational career. This consequence is assured if there are in facttwo ‘ ‘objectpositions’ ’ inEnglish: one the immediate postverbal position, and the other a positionto the right of PPcomplements.(30a) then represents the case in which all is stranded intheleftmost object position; the more degraded examples of thetype (30b) are cases in which all isstrandedin the rightmost object position. Theidea that there might be twosuch positions is not,of course, new (see especiallyKayne 1994:69–78,modifying earlier suggestions advanced in Larson1988, 1990; also Pesetsky 1989, Johnson1991, Runner 1995, Koizumi 1993, 1995, Takano 1998; see also Belletti and Shlonsky 1995for relateddiscussion). We cannot,however, follow Kayne (1994) in identifyingthis lower objectposition with the position occupied by phrasesthat (on traditional analyses) have undergone rightwardshifting. ‘ ‘Heavy’’ objectsmay appear to the right of adjuncts.

(31)She brought to the meeting at theweekend [ D P threehefty reports about local economic conditionsand prospects]. Compare(32). (32)a. What did she bring all to the meeting at the weekend? b.?What did she bring to the meeting all at the weekend? c.*What did she bring to the meeting at the weekend all? Letus take away from thisdiscussion, then, two propositions: (i) Englishhas an objectposition to the right of theposition occupied by PPcomplements butto the left of the positions occupied by adjuncts. (ii)This is apositionout ofwhichmovement to an A ¯ -positionis possible(i.e., it is a position thatis L-marked, in the terminology of Chomsky1986). Thereis much more to be saidabout this topic, and more will be said shortly. This will be more easilydone, however, once certain other observations are in place.

6.2Exceptional-Case-Marking Constructions Postal(1974) and Lasnik and Saito (1991) discuss a rangeof phenomena that suggest in combina- tionthat the exceptional-Case-marking (ECM) subject in English is in a relativelyhigh position—highenough to participate in binding and licensing phenomena involving material within adjunctsattached to thematrix VP. Thatis, the claim is that(33) and (34) have the same status. (33)a. The prosecutors proved absolutelynobody tobe guilty during any ofthetrials. b.The prosecutors proved thetwo defendants tobe guilty in each other’s trials. (34)a. The prosecutors convicted nobody during any ofthetrials. b.The prosecutors convicted thetwo defendants during each other’s trials. Lasnikand Saito (1991) argue that such cases indicate that the ECM subjectmust have raised to Q - F L O A T A N D W H -MOVEMENTINANIRISHENGLISH 69 theaccusative position associated with the matrix VP. Otherwise,it would not be sufficiently prominentto enterinto binding and licensing relations with elements inside phrases adjoined to thematrix VP. However,they remain ultimately agnostic about whether the required raising takes placeovertly or inaudibly. Lasnik (1996) returns to these questions and argues that one must assumeovert raising of the ECM subject.The core claim of the paper is that appeal to covert raisingto the accusative position is insufficient to account for thebinding and licensing facts that were dealtwith in the 1991 paper (and in Postal 1974). This is because covert A-movement is knownnot to expand binding possibilities (Den Dikken1995). 1 6 (35)a. *There seemed to themselves /eachother to beatleast two candidates well qualified for theposition. b.At leasttwo candidates seemed to themselves /eachother to be well qualified for theposition. Nor doescovert raising expand licensing configurations for negativepolarity items. (36)a. *There seemed to any of theinterviewers to be no candidateproperly qualified for theposition. b.No candidate seemed to any of the interviewers to be properly qualified for the position. If (33)is properly construed as grammatical (or, more accurately, as being as grammatical as (34)),then, the conclusion seems to be forced that raising must be overt. 1 7 Thebinding theory judgments are somewhat delicate, though. What is interesting at this pointis thatthe phenomenon of all-strandingin West Ulster English provides a morerobust way ofadjudicatingthe issue. Consider (37). (37)a. I wantedmy mother to meet lots of people at theparty. b.I expectedmy mother to meet lots of people at theparty. Thereare two ways of thinkingabout the syntax of the postverbal DP in(37): either it occupies thespecifier of thecomplement IP, orithasraised out of IPtooccupy the accusative licensing positionof thematrix clause. 1 8 Considernow what happens if we apply wh-movementto a wh-

16 There may bea contrastbetween A-andA ¯ -movementin this respect, since Fox(1995), reinterpreting some observationsof Fiengoand May (1994), has arguedthat covert A ¯ -movementdoes alter bindingpossibilities. 17 Theassumption of overtraising also (for better or forworse) removes anydoubt about whether or notan object can beproperly taken to c-command elements withinVP adjuncts.Reinhart (1976, 1983) argues in detail that this possibilitydoes exist. 18 Itis notuncontroversial that verbs like want are properlycategorized as raising /ECMverbs. However, Postal (1974:176–187), Lasnik and Freidin (1981), and Pesetsky (1982:673 –682,1991:15 –38)all presentevidence that this is thecorrect view. One mightwant to ask aboutthe status of suchuncontroversial raising /ECM verbs as believe or consider (the B- verbsof Postal 1974) in West Ulster English.These verbsin their raising /ECMuse belongto a veryformal register, however,and the task of investigatingtheir syntax in a localEnglish raises manyfraught methodological issues about register,dialect, and their interaction. I haveno confidencethat the results of that investigation would be meaningful. 70 JAMESMCCLOSKEY phrase like who all from withinan ECMcomplement.Two possibilities emerge as grammatical. (38)is unremarkable. (38)a. Who all did you want your mother to meet at the party? b.Who all did you expect your mother to meet at the party? c.Who all did you force your mother to talk to at the party? However,(39) is alsopossible. (39)a. Who did you want your mother all to meet at the party? b.Who did you expect your mother all to meet at the party? c.Who did you force your mother all to talk to at the party? And(40) is thoroughlyexcluded. 1 9 (40)a. *Who did you want all your mother to meet at the party? b.*Who did you expect all your mother to meet at the party? c.*Who did you force all your mother to talk to at theparty? Inthe case of for complements,the pattern is reversed. (41)a. Who did you arrange all for yourmother to meet at the party? b.*Who did you arrange for yourmother all to meet at the party? Thepattern in (41) is understandable in the system as developed so far. In(41a) all is stranded inthespecifier of CPheadedby for. (40b)is ungrammatical because there is no position between thespecifier of IP (occupiedby the DP your mother)andthe head of IP (occupiedby to) that couldplausibly be interpretedas a stopping-offpoint for wh-movement. What,then, should we makeof (39)and (40)? The contrast between (39)– (40) on the one handand (41) on the other suggests that the structural differences between the complement of want andthe complement of arrange mustgo beyond the presence or absenceof anovertcomple- mentizer.The pattern of contrasts falls into place right away if English (or atleast this variety ofEnglish) has overt object shift. Given this, we cantake all in(39) to mark, as usual, one of thepositions through which wh-movementhas passed (the specifier of CP).The DP your mother occursto the left of thatposition because it has undergone raising to the accusative position of

19 Thetask given to consultants to establish these conclusionswas alittledifferent from that used in general to establishthe data ofthisarticle. After itwas establishedthat a speaker hadthe basic construction,he orshe was presented with(38) and asked to supply an alternative version of the sentence alongthe lines already established (i.e., with a stranded all).Sevenspeakers were presentedwith this task, and all volunteered(39). When presented with the examples in(40), all rejected them.Examples like those in (39) are notfully grammatical forall speakers. We may attributethis tothe fact thatthey too fail tomeet theprosodic licensing requirement in the optimal way, since all isnotadjacent to a verb. Notice incidentallythat in (40) the prosodic licensing requirement is met inthe optimal way butthe structure is neverthelessstrikingly ungrammatical, suggesting once again that appeal to prosodic considerations alone is insufficient fordetermining the distribution of all strandedunder wh-movement.A similar conclusionemerges fromthe discussion ofextractionfrom postverbal subject position at (49)–(52) below. Q - F L O A T A N D W H -MOVEMENTINANIRISHENGLISH 71 thematrix clause. This interpretation further implies that English has overt verb raising. 2 0 This seemsto betheinterpretation of the data that involves the least ad hoc stipulation. If itis correct, thenat least this variety of English must have overt object shift and at least some overt verb movement.

6.3Objects Revisited If thismuch is right, then there is an obvioussolution for theproblem of thetwo object positions identifiedby stranded all. Giventhat we nowhave evidence for bothovert object shift and overt verbraising, we canidentify the leftmost (immediately postverbal) position with the VP-external accusativeposition (specifier of Agr O inmany accounts, or the outer specifier of the head that introducesexternal arguments). (42)What did you put all in the drawer? Therightmost position we canidentify with the VP-internal thematic position in which the object originates. (43)?What did you put in the drawer all? Aconsequenceof this proposal is that we mustassume that the internal organization of VP isa littledifferent from thatoften assumed. Specifically, the direct object (in these kinds of cases

20 Inaddition, movement to an A-position across aCPboundarymust be possible.There isindependentevidence, though,that this possibility exists: (a) inthe existence ofverbs in French that select thecomplementizer a` or de but exhibitraising properties (Perlmutter 1970; but cf. Kayne1981b:353, fn. 9), (i)Il commence a` pleuvoir. it begins C rain[1FIN] ‘It’s startingto rain.’ (ii)Il menace depleuvoir. itthreatens C rain[1FIN] ‘Itlooks like it might rain.’ (b)in raising across thenegative complementizer gan inIrish (McCloskey 1984, 1985, Chung and McCloskey 1987, McCloskeyand Sells 1988), (iii)N’ fhe´adfadha cuidfeola gan t abheithrighin. NEG couldits portion flesh C NEG[1FIN] be[1FIN] tough ‘Its fleshcouldn’ t butbe tough.’ and(c) inraising across for inthose English varieties thatpermit for-to complements (Henry1990, 1995). (iv)He seems forto haveleft early. As far as fundamentallocality requirements are concerned,the presence ofanintervening specifier ofCPpositionshould beirrelevantfor movement to an A-position. Furthermore, if thematic positionsare inprinciple unavailable as landing sites (Rizzi 1991,Chomsky 1995), then in the structure (v)

(v) [FAcc [VP V [CP C [IP DP I VP]]]] thespecifier ofF Acc isthe closest availablelanding site forA-movement of DP.See Pesetsky1991 for extensive related discussion.However, Norbert Hornstein (personal communication) points out that this interpretation is at oddswith Chomsky’s (1998)proposal that CP is a‘‘phase’’ thatshould block all A-movementacross itsboundary. 72 JAMESMCCLOSKEY atleast) must originate in a positionto the right of, and probably as a consequencelower than, theposition of thePP complement. 2 1 Thelower position of the object will go largely undetected, since its higher, postraising positionis the one relevantfor mostrelations that depend on relative syntactic prominence (binding ofanaphors,binding of pronouns, licensing of negativepolarity items, etc.). However, the lower originof theobject might be detectablein theintermediate status of examplessuch as (44)(Burzio 1981:337–338,nn. 4 and5).

(44)a. ?She told [ D P storiesabout each other] j tothe children tj .

b. ?We gave [D P aportraitof herself] j to the deanj .

c.??She read [ D P each otherj’sstories]to the children j . Giventhe analysis sketched above, such examples should be well formed to the limited extent thatreconstruction for PrincipleA effectsis available under A-movement. That is, their status shouldbe analogous to that of so-called backward binding examples like those in (45) (Postal 1971:188,Jackendoff 1972, Giorgi 1984, Pesetsky 1987, and especially Belletti and Rizzi 1988: 312–319).2 2 (45)a. ?Such claims about each other’ s positionsamazed them. b.?Books about themselves often annoy people. c.?*Each other’ s habitsannoy them. d.?False claims about each other’ s positionswere putabout by both candidates. e.?*Each other’ s childrenseem to them (to be) the smartest.

6.4Subjects Anumberof puzzlesarise when we askhow subjects behave under wh-quantifierfloat. We can beginwith the observation that all associatedwith an extracted subject may appear (marginally for aminorityof speakers) in postverbal position. (46)a. What happened all at theparty last night? b.Who spoke all at themeeting last night? c.Who was fightingall at theparty? d.Who was laughingall when the groom was makinghis speech? Noticethat the possibility holds both for unaccusativepredicates like happen in(46a) and for theunergative predicates seen in (46b –d).In passives also, stranding in the postverbal position ispossible. (47)a. Who was arrestedall in Duke St.? b.What was saidall at themeeting?

21 Forthese cases at least. Itis perfectlypossible that different verb types, with different thematic properties,would projectVPs withdifferent patterns of internal organization. 22 Pesetsky(1995:221 –222)makes similar observationsand draws conclusionsthat are similar inspirit,though they are developedwithin a rather differentframework ofassumptions. Q - F L O A T A N D W H -MOVEMENTINANIRISHENGLISH 73

Thelogic of precedingsections suggests that all in(47) and (46) marks either a positionin which the wh-subjectoriginates or onethrough which it haspassed. Read in thisway, (46) suggests the existenceof asubjectposition to the right of thesurface position of themain verb. Whatis this subject position? There seem to betwooptions. Either it is a rightspecifier in VP(withno implications for thescope of verbmovement in English),or itisaleftspecifier and verbmovement has applied to raise the verb to a positionto its left (Pesetsky 1989, Johnson 1991,Koizumi 1993, 1995, Runner 1995, Lasnik 1995, 1996, Bo skovic´ 1997,Takano 1998)— the latterbeing the only option available in theframework ofKayne(1994), which does not counte- nancerightward specifiers. Sincethe proposed analysis of objectpositions already commits us to theexistence of short verbraising in English, we arefree toassume that the thematic position for subjectsis a left specifier. Toestablish what exactly this position is, though, one would want to know what position the subjectoccupies in relation to other elements of theVP —objectsand PP complementsespecially. Considerobjects first. If Englishhas overt object shift and raising of V toapositionoutside VP, andif West Ulster Englishalso allows stranding of all ina VP-internalsubject position, then a straightforward predictionemerges. There should exist structures like (48), in whichthe object has raised out of VP, Vhasalso raised out of VP, and all isstrandedin the subject position inside VP.

(48) [wh-DPS u b j V DPO b j [V P [D P tS u b j all] tV tO b j (XP)]] Testingthis prediction is complicated by a numberof factors, one familiar from thepresent discussion,another from theliterature on Romance syntax. Consider the contrasts in (49)– (52). (49)a. Who all built this house? b.*Who built all this house? c.?Who built this house all? (50)a. Who all likes toffee? b.*Who likes all toffee? c.?Who likes toffee all? (51)a. Who all’ d liketea? b.*Who’ d likeall tea? c.Who’ d liketea all? (52)a. Who all was throwin’stones (around Butchers’ Gate) (yesterday)? b.*Who was throwin’all stones (around Butchers’ Gate) (yesterday)? c.Who was throwin’stones all (around Butchers’ Gate) (yesterday)? d.*Who was throwin’stones around Butchers’ Gate all yesterday? e.*Who was throwin’stones around Butchers’ Gate yesterday all? The(c) examplesare not all perfect (presumably because they do not meet the prosodic licensing requirementin the optimal way), but the contrast between the (c) examplesand the (b) examples 74 JAMESMCCLOSKEY is stark.2 3 Examplesof the (c) typeapproach total well-formedness under two conditions. First, theyimprove if the object is part of a collocationor idiom in combination with the verb —an effectvery reminiscent of the transitivity effect on postverbalsubjects in Frenchand some other Romancelanguages (Kayne 1972 and muchsubsequent work). (53c) and (54c) are close to perfect andcontrast very sharply with (53b) and (54b). 2 4 (53)a. Who all changed their mind? b.*Who changed all their mind? c.?Who changed their mind all? (54)a. Who all did their nut at the meetin’ last night? b.*Who did all their nut at the meetin’ last night? c.?Who did their nut all at the meetin’ last night? Second,they improve if the object that separates stranded all from theverb has little prosodic substance;compare (49c) with the (c) examplesof (50)– (52). If theobject is an unstressed pronoun,examples of therelevant type are essentially perfect. 2 5 (55)a. Who all read it this morning? b.*Who read all it this morning? c.Who read it all this morning? Insummary, the patterns correspond closely to expectation,as long as we factorin theprosodic concernsthat have played a rolethroughout, as well as some factors familiar (but little understood) from thestudy of postverbalsubject constructions in Romance. Finally,PP complements(in contrast with direct objects) prefer tofollow a strandedsubject all.2 6 (56)a. Who all was talkingto the kids last night? b.*Who was talkingto the kids all last night? c.*Who was talkingto him all last night? d.?Who was talkingall to the kids last night? (57)a. Who all was arguingwith the girls last night? b.*Who was arguingwith the boys all last night?

23 Notethat, although the (b) examples of(49)– (52) meet theprosodic licensing requirement in the optimal way (since all may incorporateinto V), they are thoroughlyungrammatical (suggestingonce more thatprosodic considerations donotalone determine thedistribution of stranded all; compare thediscussion of ECM cases infootnote 19). 24 Theidiom in (54)is doone’ s nut, whichmeans roughly‘ gocrazy, loseit’ . 25 Of course,(55c) is ambiguous, since, in additionto being parsed as indicatedby theunderlining, it can beparsed with it all as aconstituent—adirect object. 26 Notethat substituting prosodically light PPs (as in(56c) and (57c)) for the prosodically heavier PPs in (56b) and (57b)does not improve matters, suggestingagain that prosodic factors aloneare insufficientto account for these facts. Thisis particularlyclear whenthese facts are comparedwith the facts of(29),involving object extraction. If theproposed interpretationof (29)is right, then prosodic factors alonecannot account for the degree ofdeviance seen in(56b– c) and (57b–c). Q - F L O A T A N D W H -MOVEMENTINANIRISHENGLISH 75

c.*Who was arguingwith him all last night? d.Who was arguingall with the boys last night? Whatthese observations in combinationsuggest is an internal organization for VPlikethat schematizedin (58). The higher V of(58)is the element that introduces the external argument, andthe lower V isthe elementthat introduces internal arguments (Hale and Keyser 1993, Chomsky 1993,1995, Kratzer 1994,1996). (58) assumes an independent projection for thelicensing of accusativeDP (Agr O ),butthis is clearly not a crucialassumption. For analternative view, see Chomsky1995:348ff., 1998.

(58) FP

F AgrOP

V DPObj

AgrO VP put milk

tV tSubj

tV VP

V tObj in it

Oneimportant property of this combination of proposalsis thatit is incompatiblewith the so-calledsplit-VP hypothesis (see especiallyTravis 1992, Koizumi 1995, Harley 1995), according towhich the thematic position in which subjects originate is outside VP andis higher than the targetposition for objectshift. If theproposals here are correct, then the thematic ‘ ‘layer’’ is whollycontained within the inflectional ‘ ‘layer’’ ofthe clause, and the evidence that suggests thesplit-VP hypothesis will need to be reassessed. 2 7

27 Foradditional evidence bearing on the issue, see Jonas1996:170– 171, fns. 5 and6, McCloskey1997. See also Chomsky1998. 76 JAMESMCCLOSKEY

Itis possiblein principlethat all of these conclusions about the architecture of VP holdonly for WestUlster English and that they reveal nothing about other varieties of English. The detailed evidencewe havedrawn on hereis available only in WestUlster English. However, none of the corefacts about VP structure(the facts of constituent order, adverb placement, auxiliary placement, thesyntax of negation, ellipsis patterns, etc.) distinguish West Ulster English from, say,Standard English.It is surelyunlikely, then, that speakers of StandardEnglish and speakers of WestUlster Englishcould arrive at fundamentallydifferent grammars inthecourse of linguisticmaturation. Theformative evidence available to both groups of speakersmust be essentiallyidentical. 2 8

7 A¯ -Movement,A-Movement, and Quantifier Float Thelogic of the mechanism that allows both (1) and(3) draws nodistinction between cases in whichthe target of movementis anA ¯ -positionand cases in which the target of movementis an A-position.That being so, the same mechanism straightforwardly accounts for familiarcases of quantifierfloat, such as thoseseen in (59). (59)a. They might all have been laughing. b.They have all been laughing. c.They were alllaughing. d.They were allarrested last night. Thestructure of aphraselike they all or they both isplausibly(60).

(60) DP1

DP2

D DP they

all t2 Thatbeing so, either the lower or the higher DP willbe able to raise when attracted by some higherhead. Analysesof this general type, since they were originatedby Sportiche (1988), have been widelybut by no means universally accepted; for criticaldiscussion of the approach, and for alternativeproposals, see Klein 1976, Williams 1980, Kayne 1984:chap. 4, Dowty and Brodie 1984,Doetjes 1992, Baltin 1995, Torrego 1996, Bobaljik 1995, 1998, Morzycki 1998.

28 Iassume thatobservations about wh-quantifierfloat could not form part of the primary linguistic data. Since I firstbegan working on thetopic four or fiveyears ago,I havenot come across asingleexample inspontaneous use. Of course,this observation deepens the puzzle of howthe construction itself is acquiredby speakers ofWest Ulster English, butnot by speakers ofstandardvarieties. For tentative suggestions, see thefinal paragraph of section5. Q - F L O A T A N D W H -MOVEMENTINANIRISHENGLISH 77

Oneof thesources of skepticismabout this line of analysishas been its inability to provide anyway of understanding a well-known,but poorly understood, restriction that severely limits therange of positionsin whichthe floating quantifier may appear. Compare (59) with (61). (61)a. *They have gone all to bed. b.*They were arrestedall last night. c.*They froze all during the winter. d.*They were spokento all after class. Onthe movement analysis, it is not obvious why (61) should be impossible. Now WestUlster English is notdifferent from othervarieties of English in this.The pattern ofgrammaticalityand ungrammaticality seen in (59) and (61) is replicatedexactly in thisvariety. Thisobservation gives rise to a numberof puzzles that I wantto try to say something about in thisfinal section. We need,in the first place, an account of therestrictions seen in (61). 2 9 Tomakethe necessary distinctions,we needeither (i) anaccount that disallows an A-movement derivation for (59)and (61) and provides somealternative way of understanding(59), or (ii)an account that allows an A-movement derivation for (59),but not for (61). Sinceeither one of these approaches would be compatible with most of whatI haveargued for inthis article, and since I havenothing to add to the debate concerning (61), I willleave the matterhere. At thispoint, though, a distinctbut related puzzle presents itself— one that arises no matterwhat the correct explanation for thecontrast between (59) and (61) turns out to be. Consideran instance of strandingin the postverbal subject position. (62)Who was throwingstones all around Butchers’ Gate? As we havejust seen, stranding in this position is illegal under A-movement. (63)*They were throwingstones all around Butchers’ Gate. Itfollows from thisobservation that in (62) who mustmove directly from theVP-internal position tothe specifier of CP position.Movement of who from theVP-internal position to the specifier ofIPwouldbe indistinguishable 3 0 from themovement that gives rise to (63) and would therefore resultin ungrammaticality of the same type and of thesame source as (63) (whatever that may be).3 1

29 Theissue arises evenif some alternative(nonmovement) analysis is offeredfor grammatical instances ofquantifier float.Say we analyzedthe use of all foundin (59) as anadjoined adverbial element. Itwould still be true that the mechanisms consideredhere wouldprovide well-formed movement derivations for all oftheexamples in(59) and (61). 30 Indistinguishable,at least, at itspoint of application. I assume thatderivational ‘ ‘look-ahead’’ is impossible.Thus, there isno way toknow, at thepoint of applicationof A-movementin (62), that it will be followed by an application of wh-movementat asubsequentpoint in the derivation. 31 Thecontrast between (62)and (63) remains ifwe abandonthe movement analysis of wh-quantifierfloat. If in 78 JAMESMCCLOSKEY

From oneperspective,this conclusion is unsurprising.Movement from thepostverbal position directlyto the specifier of CPhasbeen well established for thenull subject languages since the discussioninitiated by Rizzi(1982) and the detailed supporting evidence provided subsequently byothers(see Jaeggli1982, 1985, Burzio 1986, Brandi and Cordin 1989, Kenstowicz 1989, Rizzi 1990,Poletto 1993, Campos 1997). Wh-movementof (a subpartof) thesubject in (62) is the WestUlster English equivalent of wh-movementfrom thepostverbal subject position in Italian. Morepuzzling, though, is the status of those factors (e.g., the EPP) thatwould normally makeraising of thesubject to thespecifier of TPobligatory.How areEPP (ExtendedProjection Principle)requirements satisfied in (62), if who movesdirectly to the specifier of CP, without passingthrough the specifier of TP?In minimalist terms, this is a questionabout the very intracta- bleissue of whatit means for amovement-inducingfeature to be‘‘strong.’’ Thequestion for us arisesboth for whateverfeature or features lie behind obligatory raising of the subject (EPP, subject-verbagreement, nominative licensing, etc.) and for thefeature or features that license accusativeCase. Given our earlier discussion, this feature must also be ‘‘strong,’’ forcingovert objectshift. Aplaceto start in unraveling this puzzle is with the observation that there is probably no legalderivation that includes a stepin which the interrogative pronoun moves from thespecifier position of all toan A-position. The interrogative pronoun bears the wh-feature.Its moveinto thespecifier position of all, ifnot itself triggered by the presence of a wh-featureon the head all, will,on plausible assumptions, result in that head acquiring the wh-feature(cf. Hendrick 1990,Kennedy and Merchant 1998). This will in turn mean that that position is an A ¯ -position whenoccupied by who,what, when, or where. Anyfurther movement from thisposition to an A-positionwill as a resultbe an instance of impropermovement and will be ill formed. Tomore clearly see what is atstake,consider possible derivations of thegrammatical exam- plesin (64). (64)a. Who all was throwingstones in Guildhall Square? b.Who was throwingstones all in Guildhall Square? Theinitial stages of the derivations can be schematizedas in(65).At thispoint features of Tare crucialin determining what happens next, and features within the complex DP who all are the elementswith which those features will interact. Onepossibility is that who all wouldraise to the specifier of TPandsubsequently on tothe specifierof CP,yielding(64a). 3 2

wh-quantifierfloat all is(also) taken to be anadjoined adverbial, we stillneed an account of why it may beadjoinedto theright of theobject in (62)(in the presence ofanA ¯ -trace), butnotin (63) (in the presence ofanA-trace). Theconclusion stillseems tobe forced that wh-movementmust proceed directly from the thematic positionin (62). If there were anA- trace associated with all in(62), the structure would be locallyindistinguishable from (63) and as aconsequencewould beexpectedto be ungrammatical. 32 Itshould also be possible (in all varieties ofEnglish) to strand all inone of the A-position specifiers ofthe inflectionalsystem andraise who aloneultimately to the specifier ofCP, yielding(i), Q - F L O A T A N D W H -MOVEMENTINANIRISHENGLISH 79

(65) [vP[who all] throwing stones in Guildhall Square] Object shift

[stones [vP[who all] throwing tObj in Guildhall Square]]

Short verb raising

[throwing [stones [ vP[who all] tV tObj in Guildhall Square]]]

Merge of was and raising to T

[TP was [throwing [stones [ vP[who all] tV tObj in Guildhall Square]]]] If, onthe other hand, T targets(the features of) who alone,a potentialdifficulty arises. Movement of who tothespecifier of Twouldbe illegal— an instanceof impropermovement. In addition,the impossibility of such movement is locally determinable. It must be, then, that the unavailabilityof thisderivation licenses an alternativein whichT and who enterinto an agreement relationwithout associated (visible) movement. Following introduction of C bearingthe wh- feature, who raisesto itsspecifier position and the derivation proceeds by wayof A ¯ -movement.3 3 Thepossibility of thiskind of derivational opening is expected if we adoptcertain aspects of the framework ofFrampton and Gutmann (1999), specifically if we adopt(66) (Frampton and Gut- mann1999:3). (66)A derivationconsists of asequenceof cycles,each of whichis of thefollowing form: a.(Select) A newlexical item is introduced. It selects its arguments and is merged with them. b.(Satisfy) The features of thisnewly introduced head are satisfied as fullyas possible bychecking (which induces overt movement whenever possible).

(i)Who was all throwingstones in GuildhallSquare? analogousto (ii). (ii)They were all throwingstones in Guildhall Square. However,reactions to such examples varyin ways thatI donotunderstand. Most informants I haveconsulted reject (i) andexamples likeit. Not all do,however, and two anonymous reviewers reportthat they are fullygrammatical. Ihave alsoheard at least twoexamples inactual use:(iii) from a teenage speaker ofWest Ulster English,and (iv) from a middle- agedCalifornian. (iii)Who was all down? (iv)And the VP keepsme informedof what’s all beenhappening. Ihaveno idea whatto make ofthis.Perhaps, as suggestedby areviewer, thedifficulty here has itssource in an‘ ‘agreement conundrum,’’ who forcingsingular agreement, but all suggestingplural agreement. 33 If anexpletive is includedin the initial lexical array,then of courseother, more straightforward,possibilities come into play. (i)Who was there throwingstones all downthe town? 80 JAMESMCCLOSKEY

Theessential content of (66) is the claim that movement is, in general, a preferredoption, but thatif movementshould result in a locallydetectable problem, then feature satisfaction without movementbecomes possible. In the case of (64b) ((62)), whenT isintroduced,its features must besatisfied as fully as possible. Satisfaction of thefeatures of T willdepend on features of the subjectDP, orof who inthe specifier position of the subject. If Tentersinto a checkingrelation withthe entire subject who all, movement(raising of who all)isunproblematical.On the other hand,if T entersinto a checkingrelation with the specifier who, overtmovement is impossible because,as we haveseen, it would result in a (locallydeterminable) violation: improper movement from an A¯ -toan A-position. When C issubsequently introduced, who isfree tomove overtly toits specifier, resulting ultimately in (64b) ((62)). 3 4

8Conclusion Anumberof problemsand mysteries of courseremain. But most of thefundamental properties ofthe construction we havebeen concerned with fall into place reasonably naturally given the viewof theinternal architecture of VP andthe theory of movementthat we havebeen led to.

References Aissen,Judith. 1996. Pied-piping, abstract agreement, and functional projections in Tzotzil. NaturalLan- guage& LinguisticTheory 14:447–491. Almy,Jenny. 1997. Licensing quantifiers: The w-alles constructionin German.Ms., University of California, Santa Cruz. Baker,C. L.1968.Indirect questions in English. Doctoral dissertation, University of Illinois, Urbana. Baltin,Mark. 1978. PP asa boundingnode. In Proceedingsof theEighth Annual Meeting of theNortheastern LinguisticSociety, 33–40.GLSA, Universityof Massachusetts, Amherst. Baltin,Mark. 1995. Floating quantifiers, PRO, andpredication. LinguisticInquiry 26:199–248. Beck,Sigrid. 1996. Quantified structures as barriers for LF movement. NaturalLanguage Semantics 4: 1–56. Belletti,Adriana, and Luigi Rizzi. 1988. Psych-verbs and u -theory. NaturalLanguage & LinguisticTheory 6:291–352.

34 Areviewer suggestsa connectionbetween these facts andthe celebrated English /Frenchparadigm in (i)– (iv), discoveredand discussed by Kayne(1980). (i)*I assure youChris to be the most competent. (ii)?Who can youassure me tobe competent? (iii)*Marie croit Jean eˆtre intelligent. Marie believesJean be[ 1FIN]intelligent. ‘Marie believesJean tobe intelligent.’ (iv)Jean, queMarie croite ˆtre intelligent. .. Jean C Marie believesbe[ 1FIN]intelligent. ‘Jean, whoMarie believesto be intelligent. ..’ Thewell-formedness of(ii)and (iv) in contrast to the deviance of (i)and (iii) is mysterious in that their derivations will involve(on usual assumptions) intermediate stages thatare indistinguishablein relevant respects from(i) and (iii). Somehowthe extra stepof wh-movementameliorates oreliminates whatever goeswrong in (i) and (iii). One mightthink aboutthe contrast between (62)and (63) in the same terms: A-movementto thespecifier ofTP isitself ungrammatical (hence theill-formedness of (63)),but the derivation is savedby a subsequentapplication of wh-movement(hence (62)). Q - F L O A T A N D W H -MOVEMENTINANIRISHENGLISH 81

Belletti,Adriana, and Ur Shlonsky. 1995. The order of verbalcomplements: A comparativestudy. Natural Language& LinguisticTheory 13:489–526. Bobaljik,Jonathan. 1995. Morphosyntax: The syntax of verbalinflection. Doctoral dissertation, MIT, Cam- bridge,Mass. Bobaljik,Jonathan. 1998. Floating quantifiers: Handle with care. To appear in Glot 3. Boskovic´, ZÏ eljko.1997. Coordination, object shift, and V-movement. LinguisticInquiry 28:357–365. Brandi,Luciana, and Patrizia Cordin. 1989. Two Italiandialects and the null subject parameter. In The null subjectparameter, ed.Osvaldo Jaeggli and Kenneth Safir, 111 –142.Dordrecht: Kluwer. Bresnan,Joan. 1976. On the form and functioning of transformations. LinguisticInquiry 7:3–40. Burzio,Luigi. 1981. Intransitive verbs and Italian auxiliaries. Doctoral dissertation, MIT, Cambridge, Mass. Burzio,Luigi. 1986. Italiansyntax. Dordrecht:Reidel. Campos,He ´ctor.1997. On subject extraction and the antiagreement effect in Romance. LinguisticInquiry 28:92–119. Chomsky,Noam. 1986. Barriers. Cambridge,Mass.: MIT Press. Chomsky,Noam. 1993. A minimalistprogram for linguistic theory. In Theview from Building 20: Essays inlinguistics in honor of Sylvain Bromberger, ed.Kenneth Hale and Samuel Jay Keyser, 1 –52. Cambridge,Mass.: MIT Press.[Reprinted in TheMinimalist Program, 167–217.Cambridge, Mass.: MIT Press,1995.] Chomsky,Noam. 1995. Categories and transformations. In TheMinimalist Program, 219–394.Cambridge, Mass.:MIT Press. Chomsky,Noam. 1998. Minimalist inquiries: The framework. Ms., MIT, Cambridge, Mass. To appear in Stepby step: Essays on minimalist syntax in honor of Howard Lasnik, ed.Roger Martin, David Michaels,and Juan Uriagereka. Cambridge, Mass.: MIT Press. Chung,Sandra. 1998. Thedesign of agreement:Evidence from Chamorro. Chicago:University of Chicago Press. Chung,Sandra, and James McCloskey. 1987. Government, barriers, and small clauses in Irish. Linguistic Inquiry 18:173–237. Cinque,Guglielmo. 1980. On extraction from NPs inItalian. Journalof ItalianLinguistics 5:47–99. Dikken,Marcel den. 1995. Binding, expletives, and levels. LinguisticInquiry 26:347–354. Doetjes,Jenny. 1992. Rightward floating quantifiers float to theleft. TheLinguistic Review 9:313–332. Dowty,David, and Belinda Brodie. 1984. A semanticanalysis of ‘‘floated’’ quantifiersin a transformation- lessgrammar. In Proceedingsof the Third West Coast Conference on Formal Linguistics, 75–90. Stanford,Calif.: CSLI Publications.[Distributed by CambridgeUniversity Press.] Fiengo,Robert, and Robert May. 1994. Indicesand identity. Cambridge,Mass.: MIT Press. Fox,Danny. 1995. Condition C effectsin ACD. In MITworkingpapers in linguistics27: Papers on minimalist syntax, 105–120.MITWPL, Departmentof Linguisticsand Philosophy, MIT, Cambridge, Mass. Frampton,John, and Sam Gutmann.1999. Cyclic computation: A computationallyefficient minimalist syntax. Syntax 2:1–27. Giorgi,Alessandra. 1984. Towards a theoryof long distance anaphors: A GBapproach. TheLinguistic Review 4:307–362. Giorgi,Alessandra, and GiuseppeLongobardi. 1991. Thesyntax of nounphrases: Configuration, parameters andempty categories. Cambridge:Cambridge University Press. Giusti,Giuliana. 1990. Floating quantifiers, scrambling, and configurationality. LinguisticInquiry 21: 633–641. Giusti,Giuliana. 1991. The syntax of floating alles inGerman. In Issuesin German syntax, ed. Werner Abraham,Wim Kosmeijer,and Eric Reuland, 327 –350.The Hague: Mouton. Hale,Kenneth, and Samuel Jay Keyser. 1993. On argument structure and the lexical expression of syntactic relations.In Theview from Building 20: Essays in linguistics in honor of Sylvain Bromberger, ed. KennethHale and Samuel Jay Keyser, 53 –109.Cambridge, Mass.: MIT Press. 82 JAMESMCCLOSKEY

Harley,Heidi. 1995. Subjects, events and licensing. Doctoral dissertation, MIT, Cambridge, Mass. Hendrick,Randall. 1990. Operator binding in NP. In Proceedingsof the Ninth West Coast Conference onFormal Linguistics, 249–261.Stanford, Calif.: CSLI Publications.[Distributed by Cambridge UniversityPress.] Henry,Alison. 1990. Infinitives in a for-to dialect. NaturalLanguage & LinguisticTheory 10:279–301. Henry,Alison. 1995. BelfastEnglish and StandardEnglish: Dialect variation and parametersetting. Oxford: OxfordUniversity Press. Jackendoff,Ray. 1972. Semanticinterpretation in generative grammar. Cambridge,Mass.: MIT Press. Jaeggli,Osvaldo. 1982. Topicsin Romancesyntax. Dordrecht:Foris. Jaeggli,Osvaldo. 1985. On certain ECP effectsin Spanish. Ms., University of Southern California, Los Angeles. Johnson,Kyle. 1991. Object positions. NaturalLanguage & LinguisticTheory 9:577–636. Jonas,Dianne. 1996. Clause structure, expletives and movement. In Minimalideas: Syntactic studies in the minimalistframework, ed.Werner Abraham, Samuel David Epstein, Ho ¨skuldurThra ´insson,and C. Jan-WouterZwart, 167 –188.Amsterdam: John Benjamins. Kayne,Richard. 1972. Subject inversion in French interrogatives. In Generativestudies in Romance languages, ed.Jean Casagrande and Bohdan Saciuk, 70– 126. Washington, D.C.: Georgetown University Press. Kayne,Richard. 1980. Extensions of bindingand Case-marking. LinguisticInquiry 11:75–96. Kayne,Richard. 1981a. ECP extensions. LinguisticInquiry 12:93–133. Kayne,Richard. 1981b. On certain differences between French and English. LinguisticInquiry 12:349–371. Kayne,Richard. 1984. Connectednessand binary branching. Dordrecht:Foris. Kayne,Richard. 1994. Theantisymmetry of syntax. Cambridge,Mass.: MIT Press. Kennedy,Christopher, and Jason Merchant. 1998. Attributive comparative deletion. Ms., Northwestern University,Evanston, Ill., and University of California,Santa Cruz. To appear in NaturalLanguage & LinguisticTheory. Kenstowicz,Michael. 1989. The null subject parameter in modern Arabic dialects. In Thenull subject parameter, ed.Osvaldo Jaeggli and Kenneth Safir, 263 –275.Dordrecht: Kluwer. Klein,S. 1976.A baseanalysis of the floating quantifier in French. In NELS VII:Proceedings of the SeventhAnnual Meeting of theNorth Eastern Linguistics Society. GLSA, Universityof Massachusetts, Amherst. Koizumi,Masatoshi. 1993. Object agreement phrases and the split VP hypothesis.In MIT workingpapers inlinguistics 18: Papers on Case and agreement I, 99–148.MITWPL, Departmentof Linguistics andPhilosophy, MIT, Cambridge, Mass. Koizumi,Masatoshi. 1995. Phrase structure in minimalist syntax. Doctoral dissertation, MIT, Cambridge, Mass. Koopman,Hilda. 1999. The internal and external distribution of pronominalDPs. In Beyondprinciples and parameters:Essays in memory of Osvaldo Jaeggli, ed.Kyle Johnson and Ian Roberts, 91 –132. Dordrecht:Kluwer. Kratzer,Angelika. 1994. The event argument and the semantics of voice. Ms., University of Massachusetts, Amherst. Kratzer,Angelika. 1996. Severing the external argument from its verb. In Phrasestructure and the lexicon, ed.Johan Rooryck and Laurie Zaring, 109– 137. Dordrecht: Kluwer. Larson,Richard. 1988. On the double object construction. LinguisticInquiry 19:335–391. Larson,Richard. 1990. Double objects revisited: Reply to Jackendoff. LinguisticInquiry 21:589–632. Lasnik,Howard. 1995. A noteon pseudogapping. In MITworkingpapers in linguistics 27: Papers on minimalistsyntax, 143–163. MITWPL, Departmentof Linguisticsand Philosophy, MIT, Cambridge, Mass. Q - F L O A T A N D W H -MOVEMENTINANIRISHENGLISH 83

Lasnik,Howard. 1996. Levels of representation and the elements of anaphora.Paper presented to the work- shopon Binding and Atomism, Holland Institute of General Linguistics, Leiden, February 1996. Lasnik,Howard, and Robert Freidin. 1981. Core grammar, Case theory and markedness. In Theory of markednessin generative grammar, ed.Adriana Belletti, Luciana Brandi, and Luigi Rizzi, 407 –421. ScuolaNormale Superiore, Pisa. Lasnik,Howard, and Mamoru Saito. 1991. On the subject of infinitives. In CLS 27, 324–343. Chicago LinguisticSociety, University of Chicago,Chicago, Ill. McCloskey,James. 1984. Raising, subcategorization and selection in Modern Irish. NaturalLanguage & LinguisticTheory 1:441–485. McCloskey,James. 1985. Case, movement and raising in ModernIrish. In Proceedingsof theFourth West CoastConference on Formal Linguistics, 190–205.Stanford, Calif.: CSLI Publications.[Distributed byCambridgeUniversity Press.] McCloskey,James. 1997. Subjecthood and subject positions. In Elementsof grammar, ed.Liliane Haegeman, 197–235.Dordrecht: Kluwer. McCloskey,James. 1998. The prosody of quantifierstranding under wh-movementin West Ulster English. Availablefrom http: //www.ling.ucsc.edu /, mcclosk/. McCloskey,James, and Peter Sells. 1988. Control and A-chains in Modern Irish. NaturalLanguage & LinguisticTheory 6:143–189. Merchant,Jason. 1996. Scrambling and quantifier float in German. In NELS 26, 179–193.GLSA, University ofMassachusetts,Amherst. Miyagawa,Shigeru. 1989. Structureand Case marking in Japanese. (Syntaxand Semantics 22.) San Diego, Calif.:Academic Press. Morzycki,Marcin. 1998. Floated quantifiers are all phrasal. Ms., University of Massachusetts, Amherst. Pafel, Ju¨rgen.1996. Die syntaktische und semantische Struktur von was fu¨r-Phrasen(The syntactic and semanticstructure of was fu¨r phrases). LinguistischeBerichte 161:37–67. Perlmutter,David. 1970. The two verbs begin. In Readingsin English transformational grammar, ed. RoderickJacobs and Peter Rosenbaum, 107– 119. Waltham, Mass.: Ginn. Pesetsky,David. 1982. Paths and categories. Doctoral dissertation, MIT, Cambridge, Mass. Pesetsky,David. 1987. Binding problems with experiencer verbs. LinguisticInquiry 18:126–140. Pesetsky,David. 1989. Language-particular processes and the Earliness Principle. Ms., MIT, Cambridge, Mass. Pesetsky,David. 1991. Infinitives. Ms., MIT, Cambridge, Mass. Pesetsky,David. 1995. Zerosyntax: Experiencers and cascades. Cambridge,Mass.: MIT Press. Pesetsky,David. 1998. Phrasal movement and its kin. Ms., MIT, Cambridge, Mass. Poletto,Cecilia. 1993. Lasintassi del soggetto nei dialetti italiani settentrionali (Thesyntax of the subject innorthern Italian dialects). Quaderni Patavini di Linguistica. Monografie 12. Padua: Unipress. Postal,Paul. 1971. Crossoverphenomena. New York:Holt, Rinehart and Winston. Postal,Paul. 1974. On raising. Cambridge,Mass.: MIT Press. Postal,Paul. 1997. Islands. Ms., New YorkUniversity. To appear in Thehandbook of syntactic theory, ed. MarkBaltin and Chris Collins. Oxford: Blackwell. Reinhart,Tanya. 1976. The syntactic domain of anaphora.Doctoral dissertation, MIT, Cambridge, Mass. Reinhart,Tanya. 1983. Anaphoraand semantic interpretation. Chicago:University of ChicagoPress. Reis,Marga. 1992. The category of invariant alles in wh-clauses:On syntactic quantifiers vs. quantifying particlesin German.In Whoclimbs the grammar tree?, ed.Rosemarie Tracy, 465– 492. Tu ¨bingen: MaxNiemeyer Verlag. Riemsdijk,Henk van. 1978. Acasestudy in syntactic markedness. Lisse:Peter de Ridder Press. Rizzi,Luigi. 1982. Issuesin Italiansyntax. Dordrecht:Foris. Rizzi,Luigi. 1990. RelativizedMinimality. Cambridge,Mass.: MIT Press. 84 JAMESMCCLOSKEY

Rizzi,Luigi. 1991. Argument /Adjunct(a)symmetries. Ms., Universite ´deGene`ve,SISSA, Trieste. Ross,John R. 1967.Constraints on variables in syntax. Doctoral dissertation, MIT, Cambridge, Mass. [Publishedas Infinitesyntax!, Norwood,N.J.: Ablex, 1986.] Runner,Jeffrey. 1995. Noun phrase licensing and interpretation. Doctoral dissertation, University of Massa- chusetts,Amherst. Shlonsky,Ur. 1991. Quantifiers as functional heads: A studyof quantifier float in Hebrew. Lingua 84: 159–180. Sportiche,Dominique. 1988. A theoryof floating quantifiers and its corollaries for constituent structure. LinguisticInquiry 19:425–450. Sportiche,Dominique. 1996. Clitic constructions. In Phrasestructure and the lexicon, ed.Johan Rooryck andLaurie Zaring, 213– 276. Dordrecht: Kluwer. Stowell,Tim. 1989.Subjects, specifiers, and X-bar theory. In Alternativeconceptions of phrasestructure, ed.Mark Baltin and Anthony Kroch, 232 –262.Chicago: University of ChicagoPress. Szabolcsi,Anna. 1994. The noun phrase. In Thesyntactic structure of Hungarian, ed.Ferenc Kiefer and Katalin E´ .Kiss,197 –274.(Syntax and Semantics 27.) San Diego, Calif.: Academic Press. Szabolcsi,Anna, and Frans Zwarts. 1993. Weak islands and an algebraic semantics for scope taking. Natural LanguageSemantics 1:235–284. [Reprinted in Waysof scope taking, ed.Anna Szabolcsi, 217– 262. Dordrecht:Kluwer, 1997.] Takano,Yuji. 1998. Object shift and scrambling. NaturalLanguage & LinguisticTheory 16:817–889. Torrego,Esther. 1986. Empty categories in nominals. Ms., University of Massachusetts,Boston. Torrego,Esther. 1996. On quantifier float in controlclauses. LinguisticInquiry 27:111–126. Travis,Lisa. 1992. Derived objects, inner aspect and the structure of VP. Ms.,McGill University, Montreal, Quebec. Urban,Emily. 1999. Exactly stranding.Ms., University of California,Santa Cruz. Uriagereka,Juan. 1988. On government. Doctoral dissertation, University of Connecticut, Storrs. Valois,Daniel. 1991. The internal syntax of DP. Doctoraldissertation, UCLA, LosAngeles, Calif. Williams,Edwin. 1980. Predication. LinguisticInquiry 1:203–238.

Departmentof Linguistics Universityof California, Santa Cruz SantaCruz, California 95064 [email protected]