VP-fronting Movement vs. Dislocation Dennis Ott dott@uottawa.ca University of Ottawa Feb. 26, 2016 Carleton University 1/40
The Analysis 2/40
VP-fronting in English and German Both English and German allow VPs to be fronted: (1) a. John promised to read the book, and [ VP read the book], he did. b. [ VP Das Buch gelesen] hat Peter gestern. the book read has Peter yesterday Peter read the book yesterday. I will argue that both constructions are derived in rather different ways. German VP-fronting (VPFG ) is genuine A-movement of VP. (2) [ CP [ VP das Buch gelesen] i hat Peter gestern t i ] (3) [ CP [ VP... ] i C... t i ] English VP-fronting (VPFE ) is really VP-dislocation. (4) [ VP read the book] i [ CP (that i) he did t i ] (5) [ VP... ] i [ CP κ i... t i ] (κ = correlate) 3/40
VP-dislocation elsewhere (6) [ VP... ] i [ CP κ i... t i ] VP-dislocation is a cross-linguistically widespread option: (7) a. [ VP Visitar a Ana], María suele hacer eso. visit to Ana María hab.3sg does that Visit Ana, María usually does that. b. [ VP Annat meglatogátni], azt szokta Mari Anna.acc visit that hab.3sg Mari Visit Anna, Mari usually does that. c. [ VP Anna besuchen], das tut Maria oft. Anna visit.inf that does Maria often Visit Anna, Maria does that often. (Spanish) (Hungarian) (German) The only difference: English customarily uses a null correlate. Lipták and Vicente 2009 4/40
VP-dislocation as ellipsis Basically, I m reviving Chomsky s (1977) analysis of topicalization: (8) XP i [ CP Op i... t i... ] If fronted VPs are dislocated, how do they come to show connectivity effects? (9) a. See himself i in the mirror, John i did. b. Seine i Mutter geküsst, das hat jeder Mann i schon einmal. his mother kissed that has every man already once Every man has kissed his mother at some point. (German) My solution: dislocation = deletion. (10) [ CP1 α XP i β ] [ CP2 κ i α t i β ] (11) [ CP1 John i did [ VP see himself i in the mirror] k ] [ CP2 (that k ) John did t k ] Ott 2014a, 2015 5/40
VP-dislocation as ellipsis Juxtaposition of VP-fragment and postcedent yields a coherent discourse unit. (12) [ CP1 John i did [ VP see himself i in the mirror] k ] [ CP2 (that k ) John did t k ] VP-fragments in VPF E are instances of ordinary clausal ellipsis. (13) a. What did John i do, [ VP see himself i in the mirror]? b. John i did something: [ VP see himself i in the mirror]. c. A: What did John i do? B: [ VP See himself i in the mirror]. (14) [ CP John i did [ VP see himself i in the mirror]] Parallelism Condition (following Merchant 2001, 2004) Deleted material must be semantically equivalent to a discourse-salient non-elliptical expression under focus closure. Merchant 2004; Arregi 2010; Ott and de Vries 2015 6/40
Connectivity VPF E and VPF G alike show reconstruction effects: (15) a. [ VP See himself i in the mirror], John i did. b. [ VP Sich i im Spiegel gesehen] hat Peter i. himself in.the mirror seen has Peter Peter has seen himself in the mirror. (German) In VPF G, connectivity follows from the presence of an A-chain; in VPF E, it is a side effect of clausal parallelism, requisite for deletion. (16) [ VP sich im Spiegel gesehen] hat Peter i [ VP sich i im Spiegel gesehen] (17) [ CP1 John i did [ VP see himself i in the mirror]] [ CP2 (that) John did] (18) A: What did John i do? B: See himself i in the mirror. = [ CP John i did [ VP see himself i in the mirror]] 7/40
Locality: strong islands Both VPF E and VPF G are sensitive to strong islands. (19) a. *[ VP Read the book], John doesn t know anybody who did. b. *[ VP Read the book], John regrets the fact that Mary will. (20) *[ VP Das Buch gelesen] kennt Peter niemanden der hat. the book read knows Peter no one who has Peter doesn t know anybody who has read the book. In VPF G, VP moves; in VPF E, the VP fragment s correlate does. (21) a. *[ CP VP i... [ island... t i... ]] b. [ CP1... VP i... ] *[ CP (that i )... [ island... t i... ]] However, once we turn to other islands, an asymmetry emerges. (German) 8/40
Locality: weak islands Only VPF G is rigidly constrained by weak islands. (22) a. *[ VP Das Buch gelesen] frage ich mich warum er hat. the book read wonder I refl why he has I wonder why he read the car. b. *[ VP Autos reparieren] weiß ich nicht cars repair know I not I don t know if Peter can fix cars. (23) a.?[ VP Read the book], I wonder who will. ob if b.?[ VP Fix the car], I wonder whether he will. Peter kann. Peter can VPF G patterns with predicate fronting, but VPF E with DP-fronting: (24) a. *How intelligent did you wonder whether to consider John? b.?how many books are they wondering whether to publish? Chomsky 1986; Cinque 1990 9/40
Locality: weak islands VPF G patterns with predicate fronting, but VPF E with DP-fronting: (25) *[ AP How intelligent] i did you wonder [whether to consider John t i ]? (26)?[ DP How many books] i are they wondering [whether to publish t i ]? Again, this follows if what fronts in VPF G is VP itself... (27) *[ CP [ VP Autos reparieren] i weiß ich nicht [ob Peter t i kann]] I don t know if Peter can fix cars.... whereas it is its nominal placeholder in VPF E : (28) a. [ CP1... [ VP fix the car]]?[ CP2 (that i ) I wonder [whether he will do t i ]] 10/40
Nominal-gap requirement VPF E is licensed only where the nominal pro-vp is licensed. (29) a. *Leave early, I saw them. b. [ CP1... [ VP leave early] i ] *[ CP2 (that i ) I saw them t i ] (30) a. *Get coffee, you should go. b. [ CP1... [ VP get coffee] i ] *[ CP2 (that i ) you should go t i ] VPF G is not subject to this restriction: (31) a. [ VP Tennis spielen] i solltest du öfter Tennis play should you more.often You should go play tennis more often. b. *Das i solltest du öfter t i gehen. *That you should go more often. t i gehen. go (32) [ CP [ VP Tennis spielen ] i solltest [ TP du öfter t i gehen ]] Thoms and Walkden 2015 11/40
Morphological mismatches VPF E permits morphological mismatches between the fronted VP and the host-internal gap (VPF E bleeds affixation). (33) She said she would lose her temper, and [ VP lose her temper], she has. a. *She has lose her temper. b. She has lost her temper. (34) She said she would meet me, and [ VP meet me], she has. a. *She has meet me. b. She has met me. VPF G permits no such mismatches at all, showing that VP is directly connected to its trace. (35) a. [ VP Die Fassung verloren / *verlieren] i hat sie t i. the temper lose lose has she b. Sie hat [ VP die Fassung verloren she has the temper lost She lost her temper. / *verlieren]. lose Breul 2014; Thoms and Walkden 2015; Bresnan et al. 2016 12/40
Morphological mismatches Speakers tolerate weak aspectual mismatches in short answers... (36) A: What has she done? B: Lose her temper. [ CP she did [ VP lose her temper]] (37) A: What has she done? B: Meet me. [ CP she did [ VP meet me]]... and VP-dislocation alike. (38) a.... and [ VP lose her temper], she has. b. [ CP1 she did [ VP lose her temper]] [ CP2 (that i ) she has done t i ] (39) a.... and [ VP meet me] she has. b. [ CP1 she did [ VP meet me]] [ CP2 (that i ) she has done t i ] Semantic parallelism is preserved: (40) x : she did x x : she has done x 13/40
Morphological mismatches Where mismatch is impossible under ellipsis, it is equally impossible in VPF E. (41) A: What is she doing? B: Losing her temper. B : *Lose her temper. (42) *... and [ VP lose her temper], she is. Presumably, non-parallelism is inevitable in such cases: (43) a. [ CP1 she {will/did/may} [ VP lose her temper]] b. [ CP2 (that i ) she is doing t i ] (44) x : she {will/did/may} x / x : she is doing x Parallelism Condition (following Merchant 2001, 2004) Deleted material must be semantically equivalent to a discourse-salient non-elliptical expression under focus closure. Jason Merchant, Alan Bale, p.c. 14/40
Aux-doubling VPF E permits auxiliaries to be doubled. (45) a. [ VP Willingly been examined by the committee], he certainly has been. b.?[ VP Stupidly be punished for someone else s mistakes], he probably will be. (46) [ CP1 he certainly has [ VP willingly been examined by the committee] i ] [ CP2 (that i ) he certainly has been t i ] VPF E once again patterns with other VP fragments, but VPF G never permits such auxiliary doubling. (47) A: He certainly has been WHAT? B: Willingly been examined by the committee. (48) [ VP Verurteilt worden] ist er dafür sicher convicted be is he for.that certainly He certainly has been convicted for that. (*worden). been Thoms and Walkden 2015 15/40
Remnant-VP fronting VPF E is subject to the Potential-complete VP Constraint, but VPF G is not. Potential-complete VP Constraint (Phillips 2003; Landau 2007) A fronted VP must be a potential complete VP, with the consequence that strictly subcategorized VP-material cannot be stranded. (49) a. [ VP Read the book], John did yesterday. b. *[ V(P) Read], John did the book yesterday. (50) [ V(P) Gelesen] hat Hans das Buch gestern. Hans read the book yesterday. This follows if VP in VPF E but not VPF G is a separate fragment: (51) [ CP1 John did [ VP [ VP read (the book)] yesterday]] *[ CP2 (that i ) John did t i the book yesterday] (52) [ CP1 John did [ VP [ VP read the book] yesterday]] [ CP2 (that i ) John did t i yesterday] 16/40
Remnant-VP fronting (53) a. Marion wanted to win five medals in the Olympics, and [ VP win five medals], she did. b. A: What did Marion do? B: Win five medals. (54) a. John had intended to give candy to the children on the weekend, and [ VP give candy to the children], he did. b. A: What did John intend to do on the weekend? B: Give candy to the children. (55) a. *John intended to give the children something nice to eat, and [ VP give the children], he did a generous handful of candy. b. A: What did John intend to do? B: *Give the children. (56) a. *Andrew wanted to put the lamp on the bookcase, and [ VP put the lamp], he did. b. A: What did Andrew want to do? B: *Put the lamp. Phillips 2003 17/40
Stranding of extraposed material VPF E bans stranding of extraposed material: (57) a. John said he would call people up who are from Boston, and [ VP call people up [(who are) from Boston]], he did. b. *... and [ VP call people up], he did [(who are) from Boston]. (58) *[ CP1 he did [ VP call people t k up]] [ CP2 (that i ) he did t i [(who are) from Boston] k ] VPF G does not ban such stranding: (59) Gerüchte gestreut hat er schon viele über Maria. rumors spread has he already many about Maria He already spread many rumors about Maria. (60) [ VP [ DP Gerüchte t k ] gestreut] i hat er schon viele t i [ PP über Maria] k Baltin 1981; Ott 2014b 18/40
Why (no) remnant movement? The traditional approach (den Besten and Webelhuth 1990; Müller 1998): German permits scrambling, which permits evacuation from VP. (61) [ VP t i gelesen] k hat Hans [ DP das read has Hans the Hans read the book yesterday. Buch] i book gestern t k yesterday English permits no scrambling, hence no remnant-vp fronting. Problems (Fanselow 2002; Phillips 2003; Trinh 2009): Evacuation movements have little to nothing in common with actual scrambling. Correlation doesn t seem to hold cross-linguistically: some non-scrambling languages (e.g. Hebrew, Spanish) permit RVPF. Also English: (62) a. *[ VP Persuade Bill t i], she did [to leave] i. b. [ VP Persuade her t i] though I may [to resign] i,... This approach: fronting of remnant VPs is not tied to scrambling but instead depends on the mode of VPF: movement vs. dislocation. 19/40
Prosody VP in VPF E is prone to prosodic separation; VP in VPF G isn t: (63) a. [ VP Read the book], John did. b. [ VP Das the Buch gelesen] (# ) hat Peter. book read has Peter VP in VPF E but not VPF G must bear stress, like an ellipsis fragment: (64) Q: Who read the book? A: #[ VP Read the book ], JOHN did. A : [ VP Das Buch gelesen ] hat PETER. the book read has Peter VP in VPF E is its own intonation phrase, matching its syntactic status as a root clause (= CP 1 in underlying structure). (65) a. ([ CP1... VP i... ]) IP ([ CP2... correlate i... ]) IP VPF E b. ([ CP VP i... t i ]) IP VPF G Selkirk 2011; Ott forthcoming 20/40
Conclusions VPF G is plain A-movement; VPF E is concealed dislocation. (66) [ CP [ VP das Buch gelesen] i hat Peter gestern t i ] (67) [ CP1 he did [ VP read the book] i ] [ CP2 (that i ) he did t i ] This accounts for symmetries between the two constructions... connectivity strong islands... as well as for otherwise puzzling asymmetries. weak islands morphological mismatches remnant movement intonational isolation aux-doubling These asymmetries are accounted for if VP in VPF E (unlike VP in VPF G ) is a juxtaposed, underlyingly parallel sentence fragment. 21/40
Further questions Why does English resort to dislocation rather than straightforward A-movement in VPF? Why does English use the prefield (Spec-C) so sparsely compared to, e.g., German? Are the true inversion construction those that don t allow dislocation? Is all topicalization in English covert dislocation (as per Chomsky 1977)? What s the crosslinguistic picture? Which languages have English-type VP-fronting, which are like German in permitting VPs to A-move? What about VPF with verb doubling ( predicate clefts )? Can these be analyzed as biclausal? (68) [ VP limsor et ha-mismaxim], hu masar la-memunim alav. to.hand acc the-documents he handed to-the.superiors on-him Hand the documents to his superiors, he did. (Hebrew) etc. Landau 2007 22/40
An Alternative Approach 23/40
A matching analysis of English VP-fronting (69) Eat the pies, he did. CP vp CP PRO i eat the pies Op k TP he i T did vp t k t i eat the pies Thoms and Walkden 2015 24/40
Some problems for T&W CP vp CP PRO i eat the pies Op k TP he i did t k t i eat the pies unclear how PRO gets to be (obligatorily) controlled unclear where Op is coming from, and why it is there downstairs VPE falsely predicts voice mismatches to be felicitous downstairs VPE falsely predicts vehicle change to be possible overly liberal predictions with regard to auxiliary doubling 25/40
Voice mismatches VPE (often) permits voice mismatches: (70) a. The system can be used by anyone who wants to use it. b. The janitor must remove the trash when it is apparent that it should be removed. But VPF E doesn t, contrary to T&W s predictions: (71) The system can be used by most students, a. and [ VP used by most of them], it is. b. *and [ VP used by most of them], the students do use it. (72) The janitor was told to remove the trash, a. and [ VP remove the trash], he did. b. *and [ VP remove the trash], it was removed. (73) a. [ vp PRO remove the trash], it i was [ VP removed t i ] b. [ VP used by most of them], Op k the students i do [ vp t k t i use it] Merchant 2013 26/40
Voice mismatches VPE (often) permits voice mismatches: (74) a. The system can be used by anyone who wants to use it. b. The janitor must remove the trash when it is apparent that it should be removed. Voice mismatches are never possible under clausal ellipsis: (75) a. *Joe was murdered, but we don t know who murdered him. b. *Someone murdered Joe, but we don t know who he was murdered by. Consequently, the present approach correctly rules them out. (76) a. *The janitor was told to remove the trash, and [ VP remove the trash], it was. b. *[ CP1 he did [ VP remove the trash] i ] [ CP2 (that i ) it was t i ] Merchant 2013 27/40
Vehicle change Ordinary VPE permits vehicle change (deletion of a pronoun coreferent with a discourse-given name): (77) They arrested Alex i, though he i thought they wouldn t. a. *though he i thought they wouldn t arrest Alex i. b. though he i thought they wouldn t arrest him i. But VPF E doesn t, contrary to T&W s prediction: (78) *Criticize John i, he i said I will not. (79) a. *Criticize John i, he i said I will not criticize John i b. Criticize John i, he i said I will not criticize him i Once again, the trace of VPF E doesn t behave like VPE. By not relying on VPE, my approach avoids this problem: (80) [ CP1 he i said I will not [ VP criticize John i ]] [ CP2 (that i ) he said I will not do t i ] Fiengo and May 1994 28/40
Auxiliary doubling The matching approach falsely predicts auxiliary doubling to be felicitous wherever it is under VPE. (81) a. John has [ VP willingly been examined by the committee], and Mary certainly will be. b. *[ VP Willingly been examined by the committee], Mary certainly will be. (82) [ CP [ VP willingly been examined by the commmittee] [ CP Op Mary certainly will be VP]] The dislocation approach, relying on clausal ellipsis, rules out such cases as instances of non-parallelism. (83) a. *[ CP1 Mary certainly has [ VP willingly been examined... ]] [ CP2 (that i ) Mary certainly will be t i ] b. *[ CP1 Mary certainly will be [ VP willingly been examined... ]] [ CP2 (that i ) Mary certainly will be t i ] 29/40
Obligatory control VPE tolerates sloppy identity: (84) a. John knows it s important [to [ vp PRO arb brush your teeth regularly]], and he i does [ vp t i brush his teeth regularly]. b. John i has decided [to [ vp PRO i brush his teeth regularly]], though most people k don t [ vp t k brush their teeth regularly]. Consequently, T&W over-predict possible meanings of VPF E : (85) [ vp PRO arb eat the pies] [ CP Op k he i did [ vp t k t i eat the pies]] But my approach doesn t, since it doesn t rely on Control at all: (86) [ CP1 he k did [ VP eat the pies]] [ CP2 (that i ) he k did t i ] (87) A: What did he i do? x : he i did x B: Eat the pies. (i) [ CP he i did [ VP eat the pies]] x : he i did x (ii) *[ CP he k did [ VP eat the pies]] / x : he k did x 30/40
Appendix: More on Connectivity 31/40
Connectivity Reconstruction asymmetries Cyclic connectivity in long-distance wh-movement: (88) a. [Which pictures of himself i ] did John i think that Mary saw? b. [Which pictures of himself i/k ] did John i think that Peter k saw? (89) [ CP WH did John think [ CP WH that Mary/Peter saw WH ]] Limited connectivity in long-distance VPF E : (90) a. *[ VP Criticize himself i ], John i thinks that Mary wouldn t. b. [ VP Criticize himself i/k], John i thinks that Peter k wouldn t. (91) *[ CP VP did John think [ CP VP that Mary/Peter saw VP ]] (92) *[ CP1 John i thinks that Mary wouldn t [ VP criticize himself i ]] [ CP2 (that k ) John thinks that Mary wouldn t t k ] Huang 1993 32/40
Connectivity Reconstruction asymmetries Cyclic connectivity in long-distance wh-movement: (93) a. [Which pictures of himself i ] did John i think that Mary saw? b. [Which pictures of himself i/k ] did John i think that Peter k saw? (94) [ CP WH did John think [ CP WH that Mary/Peter saw WH ]] Cyclic connectivity in long-distance VPF G : (95) [ VP Seine i/k Mutter besuchen] glaubt jeder Vater i [dass kein Junge k his mother visit thinks every father that no boy gerne tut]. gladly does Every father thinks that no boy likes to visit his mother. (96) [ CP VP... jeder Vater [ CP VP... kein Junge gerne VP tut]] 33/40
References Arregi, Karlos. 2010. Ellipsis in split questions. Natural Language & Linguistic Theory 28:539 592. Baltin, Mark. 1981. Strict bounding. In The logical problem of language acquisition, ed. Carl L. Baker and John McCarthy, 257 295. Cambridge, MA: MIT Press. den Besten, Hans, and Gert Webelhuth. 1990. Stranding. In Scrambling and barriers, ed. Günther Grewendorf and Wolfgang Sternefeld, volume 5 of Linguistik Aktuell/Linguistics Today, 77 92. Amsterdam: John Benjamins. Bresnan, Joan, Ash Asudeh, Ida Toivonen, and Stephen Wechsler. 2016. Lexical-functional syntax. Oxford: Wiley-Blackwell, 2 nd edition. Breul, Carsten. 2014. The perfect-participle paradox: some implications for the architecture of grammar. English Language and Linguistics 18:449 470. Chomsky, Noam. 1977. On WH-movement. In Formal syntax, ed. Peter W. Culicover, Thomas Wasow, and Adrian Akmajian, 71 132. San Diego, CA: Academic Press. Chomsky, Noam. 1986. Barriers. Cambridge, MA: MIT Press. Chomsky, Noam, and Howard Lasnik. 1993. The theory of principles and parameters. In Syntax, ed. Joachim Jacobs, Arnim von Stechow, Wolfgang Sternefeld, and Theo Vennemann, 506 569. Berlin/New York: de Gruyter. Cinque, Guglielmo. 1990. Types of A-bar dependencies. Cambridge, MA: MIT Press. Fanselow, Gisbert. 2002. Against remnant VP-movement. In Dimensions of movement: From features to remnants, ed. Artemis Alexiadou, Elena Anagnostopoulou, Sjef Barbiers, and Hans-Martin Gärtner, 91 125. Amsterdam: John Benjamins. Fiengo, Robert, and Robert May. 1994. Indices and identity. Cambridge, MA: MIT Press. Huang, C.-T. James. 1993. Reconstruction and the structure of VP. Linguistic Inquiry 24:103 138. Landau, Idan. 2007. Constraints on partial-vp fronting. Syntax 10:127 164. Lipták, Anikó, and Luis Vicente. 2009. Pronominal doubling under VP-topicalization. Lingua 119:650 686. Merchant, Jason. 2001. The syntax of silence. Oxford: Oxford University Press. Merchant, Jason. 2004. Fragments and ellipsis. Linguistics & Philosophy 27:661 738. Merchant, Jason. 2013. Voice and ellipsis. Linguistic Inquiry 44:77 108. Müller, Gereon. 1998. Incomplete-category fronting. Dordrecht: Kluwer. Ott, Dennis. 2014a. An ellipsis approach to Contrastive Left-dislocation. Linguistic Inquiry 45:269 303. Ott, Dennis. 2014b. Symmetric Merge and local instability: Evidence from split topics. Syntax 18:157 200. Ott, Dennis. 2015. Connectivity in left-dislocation and the composition of the left periphery. Linguistic Variation 225 290. Ott, Dennis. forthcoming. Ellipsis in appositives. Glossa. Ott, Dennis, and Volker Struckmeier. in press. Deletion in clausal ellipsis: remnants in the middle field. In Proceedings of PLC 39. Ott, Dennis, and Mark de Vries. 2015. Right-dislocation as deletion. Natural Language & Linguistic Theory (online). Phillips, Colin. 2003. Linear order and constituency. Linguistic Inquiry 34:37 90. Selkirk, Elisabeth. 2011. The syntax phonology interface. In The handbook of phonological theory, ed. John Goldsmith, Jason Riggle, and Alan C. L. Yu, 435 484. Oxford: Blackwell. Tancredi, Christopher. 1992. Deletion, deaccenting and presupposition. Doctoral Dissertation, MIT. Thoms, Gary, and George Walkden. 2015. vp-fronting with and without remnant movement. Ms., University of Glasgow and University of Manchester. Trinh, Tue. 2009. A constraint on copy deletion. Theoretical Linguistics 35:183 227. Weir, Andrew. 2014. Fragments and clausal ellipsis. Doctoral Dissertation, University of Massachusetts, Amherst. 40/40