University of Southern California Dissertations and Theses

Towards the unity of movement: implications from verb movement in Cantonese

(USC Thesis Other)

Towards the unity of movement: implications from verb movement in Cantonese

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content T O W ARDS THE UNITY O F M O VEMENT : IMP LI CA TI O NS FR O M VERB M O VEMENT IN CANT O NESE b y T omm y T sz - Ming Lee A Dissertation Pr esented to the F A CUL TY O F THE USC GRAD U A TE SCH OO L UNIVERSITY O F SO UTHERN CALIFO RNIA In P artial F ulfillment of the R equir ements f or the Degr ee DOCT O R O F P HILOSO P HY (LIN GUISTI CS ) A ugust, 2022 Cop yright 2022 T omm y T sz - Ming Lee Dedicated to my wif e Ellie, my par ents, K athy and Jimmy, and my late Gr andma. ii A CKN O WLEDGMENTS “If I ha v e seen a little further it is b y standing on the shoulders of Giants. ” – I saac N ewton (1643–1727) This dissertation w ould not exist without the “Giants” w ho I ha v e been f ortunate to w ork with during the past fiv e y ears. M y f or emost thanks goes to m y committee: A udr ey Li, Andr ew S impson, R oum y ana P anchev a, J er em y Goodman, and S tefan K eine. I must expr ess m y sincer e thanks to A udr ey Li, w ho has been encouraging and supportiv e thr ough- out m y study at USC. I am gr eatl y benefited fr om A udr ey’ s expertise in Chinese syntax, and her car eful r eading of earlier v ersions of diff er ent parts of the dissertations. H er comments ar e al w a y s critical and thought -pr o v oking, w hich push me to think wider and deeper on diff er ent topics. I t is a gr eat p leasur e to ha v e Andr ew S impson on m y committee as a co-chair . Andr ew is a sup- portiv e and caring advisor , and a fantastic linguist on Asian languages and bey ond. I learned a lot fr om him on ho w to na vigate academic lif e and handle non-academic issues. I am also inspir ed b y his passions on hiking and his lo v e to w ar ds animals. W orking with R oum y ana P anchev a has been a w onderful experience, especiall y with her detail - minded character . F r om R oumi, not onl y did I learn about syntax and semantics, but I also learned ho w to be persistent in face of linguistic puzzles and criticisms. Puzzles about quantifier scopes ar e al w a y s challenging, but I managed to pr oceed bit b y bit w hen writing up the dissertation. I am v ery glad that J er em y Goodman agr eed to be the outside member of the committee. I am impr essed b y ho w J er em y tried to understand issues that ar e bey ond his expertise, w hile still asking constructiv e questions. I also enjo y ed discussions with him in classes of p hilosop h y of languages. Last but b y no means least, I am extr emel y grateful to m y external member , S tefan K eine, with w ho I spent m y first thr ee y ears at USC. S tefan has nev er failed to surprise me on ho w he thinks about linguistic questions and ho w he understands the mechanism and the consequences of a pr oposal. Dif - f er ent ideas on head mo v ement gr ew out as term papers in S tefan ’ s classes, w hich ev entuall y become iii essential components in this dissertation. In addition to m y committee, this dissertation w ould not ha v e been possible without the peop le w ho shar e with me their judgment on Cantonese sentences. I thank Ka- W ing Chan, K enith Chan, Sheila Chan, M ei - Y ing Ki, Y ik - P o Lai, Esther Lam, Chaak - Ming Lau, M argar et Lee, T omm y Li, S um- mer M ut, Carmen T ang, Oscar W ong, and Ka- F ai Y ip, f or the time and patience at diff er ent occasions o v er the y ears. W hi le dev eloping the ideas pr esented in this dissertation, I had v arious opportunities to pr esent portions of this w ork at diff er ent v enues, w hich led to substantial impr o v ements. Earlier v ersions of this w ork has been pr esented at linguistic conf er ences, including Y ue 22, F oCaL 1, LSA 93, GLO W in Asia XIII and SI COGG XXI, W CCFL 38, N A CCL 32, and NELS 51, as w ell as at r eading gr oups/ seminars at S tanf or d U niv ersity , U niv ersity of Calif ornia, Los Angeles, U univ ersity Connecticut, and Y ale U ni - v ersity . I thank the audience at the abo v e occasions. I must also thank the peop le w ho selflessl y shar ed their ideas with me. I w ould lik e to ackno w ledge, to the best of m y memory , Željk o Bošk o vić, K en y on B ranan, La wr ence Cheung, Colin Da vis, Michael Y oshitaka Erlewine, R obert F rank, V era Gribano v a, Boris H arizano v , H ajime H oji, Khalil I skar ous, P aul La w , W infried Lechner , P eppina P o- lun Lee, T ra vis M ajor , V ictor J unnan P an, Luis Miguel T oquer o P ér ez, Da vid P esetsky , E than P oole, I an R oberts, Deniz R udin, Barry Schein, S ze- W ing T ang, H aley W ei W ei, Alexis W ell w ood, Ka- F ai Y ip, and Raffaella Zanuttini. This list is pr obabl y incomp lete. I apologize to the peop le w ho I ha v e talk ed to but ar e not on the abo v e list. M an y peop le ha v e made m y lif e in graduate school w a y mor e enjo y able. I ha v e to thank the mem- bers of m y cohort, S il via Kim, Y ijing Lu, and J ina Song, w ho al w a y s br ought me to good K or ean f ood and desserts. W e do not o v erlap much on academic pursuits but I r eall y enjo y ed talking to y ou all. I also thank the v ery lo v el y and energetic USC f olks, Samir Alam, Betül Erbaşı, Elango K umaran, Sarah H y e- y eon Lee, Miran Or , Luis Miguel T oquer o P ér ez, Daniel Plesniak, J esse S torbeck, H aley W ei W ei, A dam W oodnutt, Y ifan Y ang, and Y ubin Zhang. Special thanks go to Guillermo R uiz, our administra- tiv e expert in the department, w ho handled tones of inquiries and emails. iv I also w ant to thank the members of the Cantonese/ Chinese online r eading gr oups during the pandemic, w hich k ept me compan y thr oughout the difficult times ( although w e ar e in diff er ent parts of the w orld): Zhuo Chen, J iahui H uang, Ka- W ing Chan, Sheila Chan, Esther Lam, M argar et Lee, Minqi Liu, J ia R en, Carmen T ang, H uilei W ang, Bo X ue, and Ka- F ai Y ip. M y lif e w ould ha v e been substantiall y diff er ent if I ha v e not app lied f or Ph.D . pr ogram after m y master study . I o w e a big thank - y ou to Y uki T akubo, S ze- W ing T ang, and M ag dalena Kaufmann f or encouraging me to further m y academic pursuits. Along this journey , I w ould lik e to thank Luis Miguel T oquer o P ér ez, and Ka- F ai Y ip once again f or both academic and non-academic supports. I wish I could off er mor e than I took fr om them o v er the y ears. I am indebted to m y famil y f or all their lo v e and patience, and f or allo wing me to do w hat I w ant. I am extr emel y grateful that they al w a y s w elcome me home with big dinners, tell me their funn y stories o v er the past months, and driv e me to the airport ev ery single time I lea v e H ong K ong. Let me tak e this opportunity to thank m y late grandma, w hose influence on me nev er ceases to exist. F or m y w if e Ellie, it is all bey ond w or ds. v T ab l e of Contents Dedi cati on ii A ckno w l edgments iii List of T ab l es xi A bbr e viati ons xii A bstr act xiii Pr eface xi v 1 Intr oducti on 1 1.1 The theor etical goals and emp irical domains of the thesis . . . . . . . . . . . . . . . . 1 1.2 A unity appr oach to mo v ement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.3 The outline of this thesi s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2 A ppr oaching head mo v ement 8 2.1 Intr oduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.2 The origin of head mo v ement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.2.1 F r om independent transf ormation rules to M o v e- . . . . . . . . . . . . . . 8 2.2.2 Constraints on head mo v ement . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.3 The debates surr ounding he ad mo v ement . . . . . . . . . . . . . . . . . . . . . . . . 13 2.3.1 Theor etical concerns of the adjunction appr oach to head mo v ement . . . . . 13 2.3.1.1 The Extension Condition . . . . . . . . . . . . . . . . . . . . . . . 14 2.3.1.2 The Empty Category Princip le/ Pr oper B inding Princip le . . . . . 15 2.3.1.3 The non-successiv e cy clic natur e . . . . . . . . . . . . . . . . . . 16 2.3.1.4 Locality constraints . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.3.1.5 The Chain U nif ormity Condition . . . . . . . . . . . . . . . . . . 18 2.3.2 Empirical diff er ences with p hrasal mo v ement? . . . . . . . . . . . . . . . . . 19 2.3.2.1 The locality constraints on head mo v ement . . . . . . . . . . . . . 19 2.3.2.2 The interpr etiv e eff ects of head mo v ement . . . . . . . . . . . . . 20 2.3.2.3 The morp ho-p honological r ealization of head mo v ement . . . . . 21 2.4 N on-unity appr oaches to head and p hrasal mo v ement . . . . . . . . . . . . . . . . . 22 vi 2.4.1 Eliminating head mo v ement fr om the syntax . . . . . . . . . . . . . . . . . . 22 2.4.1.1 P ost -syntactic mo v ement/ operations . . . . . . . . . . . . . . . . 22 2.4.1.2 R emnant p hrasal mo v ement . . . . . . . . . . . . . . . . . . . . . 23 2.4.2 R ef ormulating head mo v ement in the syntax . . . . . . . . . . . . . . . . . . 24 2.4.3 Interim summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 2.5 R ecent pursuits of a uni fied theory of mo v ement . . . . . . . . . . . . . . . . . . . . 25 2.5.1 H ead mo v ement to the specifier position . . . . . . . . . . . . . . . . . . . . 26 2.5.2 N o head-specific locality constraint . . . . . . . . . . . . . . . . . . . . . . . 27 2.5.3 The interpr etation of head mo v ement . . . . . . . . . . . . . . . . . . . . . . 28 2.6 S ummary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 3 Interv enti on eff ects: v erb mo v ement to perip her al positi ons 29 3.1 Intr oduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 3.2 Interv ention eff ects and he ad mo v ement . . . . . . . . . . . . . . . . . . . . . . . . . 34 3.2.1 Interv ention due to identical structural type s . . . . . . . . . . . . . . . . . . 35 3.2.2 The particular natur e of the HM C and ex ceptions to t he HM C . . . . . . . . 37 3.2.3 Base generation and r emnant mo v ement and as alternativ es . . . . . . . . . 42 3.2.4 Interim summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 3.3 V erb doubling constructions and discourse eff ects . . . . . . . . . . . . . . . . . . . 45 3.3.1 T ypes of v erbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 3.3.2 M orp ho-syntactic pr operties and v ariants . . . . . . . . . . . . . . . . . . . 52 3.3.2.1 T opic constructions of v erbs . . . . . . . . . . . . . . . . . . . . . 52 3.3.2.2 ‘E v en ’ - f ocus constructions of v erbs . . . . . . . . . . . . . . . . . 54 3.3.2.3 Copula f ocus constructions of v erbs . . . . . . . . . . . . . . . . . 55 3.3.2.4 Dislocation cop ying of v erbs . . . . . . . . . . . . . . . . . . . . . 57 3.3.2.5 Interim summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 3.3.3 Discourse eff ects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 3.3.3.1 Contrastiv e v erbal topics . . . . . . . . . . . . . . . . . . . . . . . 59 3.3.3.2 A dditiv e v erbal f oci . . . . . . . . . . . . . . . . . . . . . . . . . . 60 3.3.3.3 Exhaustiv e v erbal f oci . . . . . . . . . . . . . . . . . . . . . . . . 62 3.3.3.4 Def ocused v erbs . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 3.3.3.5 Interim summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 3.4 E vidence f or v erb mo v ement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 3.4.1 Lexical identity eff ects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 3.4.2 I sland eff ects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 3.4.2.1 I sland sensitivity . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 3.4.2.2 Long -distance/ Cr oss-clausal dependencies . . . . . . . . . . . . . 76 3.4.2.3 Interim summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 3.4.3 I diomatic expr essions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 3.5 F ocus Interv ention Eff ects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 3.5.1 N o interv ention b y heads . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 3.5.2 Interv ention b y f ocused elements . . . . . . . . . . . . . . . . . . . . . . . . 81 3.5.3 N o interv ention b y quantificational elements . . . . . . . . . . . . . . . . . 89 3.5.4 Interim summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 3.6 Pr oposal: head mo v ement to the specifier position . . . . . . . . . . . . . . . . . . . 92 3.6.1 Details of the pr oposal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 3.6.2 An illustration of the pr oposal . . . . . . . . . . . . . . . . . . . . . . . . . . 95 3.6.3 Deriving the pr operties of v erb doubling const ructions . . . . . . . . . . . . 98 3.6.3.1 The or dering of the functional pr ojections in the CP perip hery . . 98 3.6.3.2 The mo v ement pr operties in v erb doubling constructions . . . . . 99 3.6.3.3 A syntactic exp lanation to F ocus Interv ention Eff ects . . . . . . . 100 3.7 Alternativ e anal y ses to a he ad mo v ement appr oach . . . . . . . . . . . . . . . . . . . 103 3.7.1 N on- mo v ement appr oaches . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 3.7.1.1 Base generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 3.7.1.2 Base generation p lus operator mo v ement . . . . . . . . . . . . . . 106 3.7.2 Phrasal mo v ement appr oaches . . . . . . . . . . . . . . . . . . . . . . . . . 108 3.7.2.1 R emnant VP mo v ement . . . . . . . . . . . . . . . . . . . . . . . 108 3.7.2.2 VP mo v ement with subsequent deletion . . . . . . . . . . . . . . 110 3.8 Discussions and imp lications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 3.8.1 R ef ormulating the H ead M o v ement Constraint . . . . . . . . . . . . . . . . 112 3.8.2 A parallel anal y sis with p hrasal mo v ement . . . . . . . . . . . . . . . . . . . 114 3.8.3 F ocus Interv ention Eff ects in p hrasal mo v ement . . . . . . . . . . . . . . . . 117 3.9 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 4 Scope eff ects: mo v ement of quantifi cati onal heads 121 4.1 Intr oduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 4.2 (N on-)occurr ence of semantic eff ects with head mo v ement: an ongoing debate . . . 124 4.2.1 A lack of semantic eff ects? . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 4.2.2 Discourse eff ects of head mo v ement . . . . . . . . . . . . . . . . . . . . . . 129 4.2.3 Scope eff ects of head mo v ement . . . . . . . . . . . . . . . . . . . . . . . . . 129 4.2.3.1 M o v ement of (quantificational) determiners . . . . . . . . . . . . 130 4.2.3.2 M o v ement of negation . . . . . . . . . . . . . . . . . . . . . . . . 132 4.2.3.3 M o v ement of modal v erbs . . . . . . . . . . . . . . . . . . . . . . 137 4.2.3.4 M o v ement of aspectual v erbs . . . . . . . . . . . . . . . . . . . . 139 4.2.4 Interim summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 4.3 The distribution of aspect ual v erbs and modal v erbs . . . . . . . . . . . . . . . . . . 141 4.3.1 The ( r estricted) high position . . . . . . . . . . . . . . . . . . . . . . . . . . 141 4.3.2 V erbs that can appear in the high position . . . . . . . . . . . . . . . . . . . 143 4.3.2.1 Aspectual v erbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 4.3.2.2 M odal v erbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 4.3.2.3 Interim summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 4.3.3 Quantificational elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 4.3.4 F ocused elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 4.3.5 Interim summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 4.4 Pr oposal: scope-shifting head mo v ement . . . . . . . . . . . . . . . . . . . . . . . . 156 4.4.1 T w o components of the pr oposal . . . . . . . . . . . . . . . . . . . . . . . . 156 4.4.1.1 Ov ert scope-shifting head mo v ement . . . . . . . . . . . . . . . . 157 4.4.1.2 Scope Econom y . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159 4.4.2 Deriving the pr operties of mo v ement of quantificat ional heads . . . . . . . . 162 4.4.2.1 Deriving the quantificational scope eff ects . . . . . . . . . . . . . 162 4.4.2.2 Deriving the f ocus scope eff ects . . . . . . . . . . . . . . . . . . . 165 4.4.2.3 Deriving the r estriction on v erbs . . . . . . . . . . . . . . . . . . 168 4.4.3 R emarks on the landing site and the trigger . . . . . . . . . . . . . . . . . . 172 4.4.3.1 The landing site . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172 4.4.3.2 The trigger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 4.5 F urther evidence f or the p r oposal . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177 4.5.1 S tacking of quantificational heads in the high position . . . . . . . . . . . . 178 4.5.2 Shortest M o v e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 4.5.3 M o v ement out of coor dinate structur es . . . . . . . . . . . . . . . . . . . . . 183 4.5.4 A r emark on the indeterminacy of island sensitivity . . . . . . . . . . . . . . 185 4.6 Alternativ e anal y ses to a he ad mo v ement appr oach . . . . . . . . . . . . . . . . . . . 187 4.6.1 M ultip le base positions of aspectual v erbs and modal v erbs . . . . . . . . . . 188 4.6.2 An in-situ appr oach to aspectual v erbs and modal v erbs . . . . . . . . . . . . 190 4.6.3 A r emnant mo v ement appr oach . . . . . . . . . . . . . . . . . . . . . . . . . 191 4.6.4 M o v ement of aspectual v erbs and modal v erbs as p hrasal mo v ement . . . . . 193 4.7 Discussions and imp lications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194 4.7.1 Semantic eff ects of head mo v ement . . . . . . . . . . . . . . . . . . . . . . . 194 4.7.2 A parallel observ ation with p hrasal elements . . . . . . . . . . . . . . . . . . 195 4.7.3 The trigger of head mo v ement . . . . . . . . . . . . . . . . . . . . . . . . . . 198 4.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 5 Linearizati on: doub ling eff ects of heads and p hr ases 203 5.1 Intr oduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204 5.2 Asymmetries in doubling in Can tonese . . . . . . . . . . . . . . . . . . . . . . . . . 207 5.3 Pr oposal: C y clic Linearization and Cop y Deletion suspension . . . . . . . . . . . . . 210 5.3.1 C y clic Linearization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210 5.3.2 Cop y Deletion suspension . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212 5.4 Deriving the asymmetries in doubling . . . . . . . . . . . . . . . . . . . . . . . . . . 215 5.4.1 Licit and illicit cases in topic constructions . . . . . . . . . . . . . . . . . . . 216 5.4.2 Licit, illicit and optional cases in right dislocati on . . . . . . . . . . . . . . . 219 5.4.2.1 Licit cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219 5.4.2.2 Illicit cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221 5.4.2.3 Optional cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223 5.4.3 A r emark on diff er ences in acceptability . . . . . . . . . . . . . . . . . . . . 228 5.4.4 R esol ving a further asymmetry in doubling . . . . . . . . . . . . . . . . . . . 229 5.5 Alternativ e exp lanations t o the doubling eff ects . . . . . . . . . . . . . . . . . . . . . 231 5.6 Extension: v erb mo v ement without doubling . . . . . . . . . . . . . . . . . . . . . . 233 5.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240 6 Conclusi ons 242 R ef er ences 243 List of T ab l es 2.1 S ummary of evidence of scope e ff ects with head mo v ement . . . . . . . . . . . . . . 21 3.1 The w or d or der patterns illu strated in (27) and (28) . . . . . . . . . . . . . . . . . . . 33 3.2 The schematic pattern of topic con structions of v erbs . . . . . . . . . . . . . . . . . 53 3.3 The schematic pattern of ‘ ev en ’ - f ocus constructions of v erbs . . . . . . . . . . . . . . 54 3.4 The schematic pattern of copul a f ocus constructions of v erbs . . . . . . . . . . . . . 56 3.5 The schematic pattern of dislo cation cop ying of v erbs . . . . . . . . . . . . . . . . . 58 3.6 The schematic patterns of v erb doubling constructions . . . . . . . . . . . . . . . . . 58 3.7 The discourse eff ects of t he v erb doubling constructions . . . . . . . . . . . . . . . . 68 3.8 The interv ention eff ects o bserv ed with v erb doubling constructions . . . . . . . . . . 87 3.9 F eatur e specification of the f ocus and def ocus f eatur es in Cantonese . . . . . . . . . 93 3.10 Distribution of the unint erpr etable f ocus/ def ocus f eatur es and their r ealizations . . . 94 3.11 The discourse eff ects of t he v erb doubling constructions ( r epeated) . . . . . . . . . . 98 3.12 The w or d or der patterns illu strated in (27), (28) and (146) . . . . . . . . . . . . . . . 115 4.1 S ummary of evidence of scope e ff ects with head mo v ement . . . . . . . . . . . . . . 130 4.2 T w o scenarios of exam r esults i n a class of thr ee . . . . . . . . . . . . . . . . . . . . 142 4.3 A parallel quantificational ana l y sis of aspectual v erbs and modal v erbs . . . . . . . . 159 5.1 Doubling asymmetries in Canto nese . . . . . . . . . . . . . . . . . . . . . . . . . . . 209 5.2 V erb doubling and r emnant mo v ement acr oss languages . . . . . . . . . . . . . . . . 240 xi A bbr e viati ons a cc accusativ e ( case ) cl classifier co n contrastiv e mark er co nt continuativ e ( aspect ) co p copula v erb d a t dativ e ( case ) d ecl declarativ e d el delimitiv e ( aspect ) d is disposal mark er exp experiential ( aspect ) foc f ocus mark er imp imperf ectiv e ( aspect ) inf infinitiv e masc masculine (gender ) m o d modifier mark er neg negation n o m nominativ e ( case ) pst past ( tense ) perf perf ectiv e ( aspect ) pr og pr ogr essiv e ( aspect ) q question p article rel r elativ e mar k er res r esultativ e sfp sentence - final particle sg singular to p topic mark er xii ABSTRA CT T O W ARDS THE UNITY O F M O VEMENT : IMPLI CA TI O N S FR O M VERB M O VEMENT IN CANT O NESE T omm y T sz - Ming Lee Disp lacement ( of linguistic expr essions ) is a ubiquitous p henomenon in natural language. In the generativ e tradition, disp lacement is modeled in terms of tr ansf ormation , or mor e pr ecisel y , move - ment , w hich establishes dependencies among syntactic constituents in a p hrase structur e. This thesis pr obes the question r egar ding to w hat extent mo v ement theories can be unified. Specificall y , I ad- dr ess issues surr ounding the debate of the distinction betw een head movement and phr asal movement o v er the past f ew decades. The distinction pr esupposes that structural comp lexity of the mo ving el - ement is corr elated with its mo v ement pr operties. The goal of this thesis is to sho w that this is an un w arranted assumption. Based on a number of case studies on v erb disp lacement p henomena in Cantonese, I attempt a unified theory of mo v ement b y abandoning the head/ p hrase distinction in mo v ement theories. P articularl y , I sho w (i) that v erbs in Cantonese can undergo syntactic mo v e- ment to the perip heral position of a sentence and is subject to general locality / minimality constraints on mo v ement, and (ii) that their mo v ement ma y aff ect semantic interpr etation, leading to discourse eff ects and scope eff ects that ar e commonl y observ ed in p hrasal mo v ement. I further argue, with evi - dence fr om linearization, that head mo v ement and p hrasal mo v ement in Cantonese ar e subject to the same mechanism w hen determining the pr onunciation of the mo v ement chains. These observ ations con v erge on the conclusion that the p hrase structur e status of syntactic constituents bears a minimal r ole in theorizing disp lacement p henomena in natural language. This thesis r epr esents a minimalist pursuit of a unified theory of mo v ement. xiii Pr e face Chapter 1 sets up the r esear ch question and establishes the major arguments in this thesis. The cen- tral idea is that mo v ement operations do not mak e r ef er ence to p hrase structural diff er ences betw een heads and p hrases. The empirical evidence comes fr om v arious cases of v erb disp lacement in Can- tonese. Chapter 2 traces the origin and dev elopment of the notion of head mo v ement since the 1970s. W hile the notion of head mo v ement has pr o v ed empiricall y useful in capturing v arious linguistic p henomena, it also led to debates r elating to theor etical and empirical issues since the earl y minimalist period. I r eview r ecent r esponses to the issues surr ounding head mo v ement. Chapter 3 examines potential interv ening elements in head mo v ement. The discussion builds on f our v erb doubling constructions that come with ( diff er ent ) discourse eff ects. I t is first argued that the v erbs in these constructions undergo mo v ement to the specifier position of a functional head in the left perip hery . I t is further sho wn that, w hile a head does not block the v erb mo v ement, a f ocused element ma y lead to interv ention (i.e., F ocus Interv ention Eff ects ). This pr operty is argued to f ollo w fr om a minimality condition of the operation Agr ee that mak es r ef er ence to syntactic f eatur es (Chomsky 2000 , 2001 ). The findings r ev eal that the H ead M o v ement Constraint does not app l y to all instances of head mo v ement, and that syntactic interv ention eff ects ar e observ ed with head mo v ement, on a par with p hrasal mo v ement. Chapter 4 diagnoses an instance of head mo v ement that induces scope eff ects. I argue that quan- tificational heads such as aspectual v erbs and ( a subset of ) modal v erbs in Cantonese can undergo ( o v ert ) head mo v ement to achiev e scope enrichment. F urthermor e, this mo v ement is constrained b y an econom y condition, Scope Econom y , w hich is independentl y observ ed with mo v ement of p hrasal xiv quantifiers (F o x 2000 ). The findings suggest that head mo v ement is no diff er ent fr om p hrasal mo v e- ment in terms of the potentials to induce semantic eff ects, and that Scope Econom y constrains both head and p hrasal mo v ement. Chapter 5 discusses the issue of ho w mo v ement chains of heads ar e pr onounced and linearized. I t concerns the doubling eff ects of head and p hrasal mo v ement in Cantonese. Empirical data r ev eal that the doubling eff ects ar e not specific to mo ving heads and that head mo v ement does not al w a y s lead to doubling eff ects. I t is suggested that doubling eff ects arise fr om the fact that the operation r esponsible f or erasing copies in a mo v ement chain is r egulated b y p honological r equir ements that f ollo w fr om a v ersion of C y clic Linearization (F o x and P esetsky 2005 ). S uch an account deriv es the Cantonese doubling pattern of heads and p hrases without r ecourse to the p hrase structur e status of the ( non-)doubling elements. I t maintains that the mechanism that determines cop y pr onunciation is the same f or head chains and p hrase chains. Chapter 6 concludes the thesis. xv Chapter 1 Intr oducti on 1.1 The theor eti cal goals and empiri cal domains of the thesis The theor etical goal of this thesis is to contribute to our understanding of mo v ement theories of nat - ural language. I pick up ongoing debates of the theor etical status and empirical pr operties of head mo v ement, and exp lor e the possibility of a unified theory of mo v ement that does not mak e r ef er ence to structural types such as heads and p hrases. The distinction betw een head mo v ement and p hrasal mo v ement pr esupposes that structural comp lexity of the mo ving element is corr elated with its mo v e- ment pr operties. I argue that this is an un w arranted assumption. Specificall y , I argue that the r ole of the head-p hrase distinction is minimal in mo v ement theories: both types of constituents ar e targeted b y the same mo v ement operation. S upporting evidence comes fr om observ ations that mo v ement of heads and p hrases ar e subject to the same set of syntactic princip les, w hich constrain (i) ho w they mo v e in the syntax, (ii) ho w they contribute to interpr etation, and (iii) ho w their chains ar e p honolog - icall y r ealized. T o the extent that head mo v ement can be assimilated to p hrasal mo v ement, this thesis sets the basis of a mo v ement theory that does not discriminate heads fr om p hrases, hence a unified theory of mo v ement. The empirical domains of this thesis ar e constituted b y a number of v erb disp lacement cases in (H ong K ong) Cantonese. The first type concerns w hat I r ef er to as verb doubling constructions , w her e an additional cop y of a v erb appears in the (left or right ) perip heral position of a sentence, and con- v ey s diff er ent discourse eff ects ( e.g. topic - or f ocus- r elated interpr etations ). These constructions ar e 1 1.1. THE THEO RETI CAL GO ALS AND EMP IRI CAL DO MAINS O F THE THESIS ex emp lified in ( 1 ). The sentences in ( 1 a-c ) ar e often r egar ded as pr edicate cleft constructions. The anal ytical questions posited b y these constructions concern the deriv ation of these sentences and the r elationship betw een the v erb in the base position and the v erb in the perip heral position. These issues ar e discussed in details in Chapter 3 and Chapter 5. (1) Examp les of v erb doubling constructions a. V S V O: T opic constructions of v erbs ( cf. Cheng and V icente 2013 , p.13) M aai buy k eoi s/he hai co p maai -gw o buy -exp go-bun that -cl syu. book ‘ As f or buying, s/he has bought that book (but...). ’ b. Lin - V S V O: ‘E v en ’ - f ocus constru ctions of v erbs ( cf. Cheng and V icente 2013 , p.2) Lin ev en tai r ead k eoi s/he dou also m- tai not - r ead ni -bun this-cl syu. book ‘S/he didn ’t ev en READ this book. ’ c. H ai - V S V O: Copula f ocus constructions of v erbs H ai co p dim touch Aaming Aaming m-gam not -dar e dim touch ni - zek this-cl dungmat animal ze1. sfp ‘ Aaming dar e not to T O U CH this animal onl y . ’ d. S V O sfp V : Right dislocation/ dislocation cop ying of v erbs (K. K. Chan 2016 , p .18, adapated) Zoengsaam Zoengsaam gammaan tonight fan sleep ni - zoeng this-cl cong bed aa3 sfp fan . sleep ‘Zoengsaam ( will) sleep on this bed tonight. ’ Another type of v erb disp lacement discussed in this thesis concerns aspectual v erbs and a subset of modal v erbs. In the sentences in ( 2 ), hoici ‘begin ’ can appear a lo w position (f ollo wing the subject ) or a high position (pr eceding the subject ). The tw o sentences con v ey diff er ent scope interpr etations. They beg the question of w hether the tw o sentences ar e deriv ationall y r elated, and w hat ar e the r elativ e 2 1.2. A UNITY AP P R O A CH T O M O VEMENT constraints on the alternation of w or d or der . These sentences ar e examined in depth in Chapter 4. (2) The lo w and high positions of houci ‘begin ’ a. ‘ onl y’ > ‘begin ’ / *‘begin ’ > ‘ onl y’ Dak onl y Aaming Aaming hoi ci begin haau -dou get -able hou good singzik. r esult ‘Onl y Aaming is such that he begins to get good r esults. ’ b. *‘ onl y’ > ‘begin ’ / ‘begin ’ > ‘ onl y’ H oi ci begin dak onl y Aaming Aaming haau -dou get -able hou good singzik. r esult ‘I t begins to be the case that onl y Aaming is getting good r esults. ’ As a note on the methodology , the judgment of the Cantonese sentences thr oughout this paper come fr om fiv e (H ong K ong) Cantonese speak ers ( unless specified otherwise ). Sentences without marking or mark ed b y O K sho w that all the speak ers find the sentences acceptable. An asterisk sym- bol * indicates that all/ most speak ers find the sentences unacceptable, w her eas question marks ?/?? indicate that the speak ers find the examp le unnatural or degraded, but not entir el y unacceptable. 1.2 A unity appr oach to mo v ement In the generativ e tradition, p henomena r elating to disp lacement of linguistic elements ar e modeled as (independent ) transf ormational rules, or mo v ement operations. In the course of theorizing dis- p lacement p henomena, ther e is a constant tension betw een the theor etical desir e f or unification and empirical challenges r ooted fr om the div erse natur e of the p henomena r elating to head mo v ement. On one hand, the desir e f or unification underlies the eff orts of r educing ( diff er ent, construction- specific ) transf ormational rules that app l y to v erbs to a single syntactic pr ocess. This giv es rise to the first characterization of H ead M ovement (i.e., the adjunction appr oach to head mo v ement ) in the 1980s, in w orks b y K oopman ( 1984 ), T ra vis ( 1984 ), and Bak er ( 1988 ), among others. H ead M o v ement un- der the Go v ernment and B inding F ramew ork (Chomsky 1981 , 1986 ) can be further subsumed under 3 1.2. A UNITY AP P R O A CH T O M O VEMENT the notion of M ove - (i.e., mo v e an ything an yw her e ), w hich r epr esents the most unspecified f orm of mo v ement operation ( and it also app lies to p hrases ). In the later minimalist framew ork, mo v ement is modeled as a sub-type of the operation M erge , namel y , I nternal M erge (Chomsky 1995b , 2000 , 2001 ), w hich, again does not distinguish heads fr om p hrases in terms of mo v ement. H o w ev er , on the other hand, the attempts to unification face both theor etical and empirical chal - lenges. 1 H ead mo v ement, or mor e pr ecisel y , the adjunction appr oach to head mo v ement, violates a number of syntactic princip les that mo v ement operations ar e expected to obey , including the Exten- sion Condition, the Empty Category Princip le, and so on. A dditionall y , head mo v ement appears to be qualitativ el y diff er ent fr om p hrasal mo v ement in terms of empirical pr operties. F or examp le, head mo v ement is said to obey a stricter locality constraint (i.e., the H ead M o v ement Constraint, T ra vis ( 1984 )) and it is also said to fail to induce semantic eff ects in the same w a y as p hrasal mo v ement ( e.g., Chomsky 2000 ). These diff er ences in vite pr oposals that adopt a non-unity appr oach to mo v ement, w her e head mo v ement is substantiall y r ef ormulated in a w a y that departs fr om the mechanism held r esponsible f or p hrasal mo v ement. In spite of the div erse natur e of the p henomena concerning head mo v ement, I suggest that it does not necessaril y r eflect the non-unif orm natur e of movement oper ations in syntax. The div erse natur e of the head mo v ement p henomena ma y be attributed to the fact that the disp lacement pr operties of heads ar e r esulted fr om diff er ent operations in diff er ent components of the grammar . This thesis puts itself along the line of ( the continuation of ) the minimalist pursuit of a mo v ement theory that does not diff er entiate heads fr om p hrases, w hich, to diff er ent extents, underlines the spirit in w orks b y F ukui and T akano ( 1998 ), T o y oshima ( 2000 , 2001 ), T akahashi ( 2002 ), M atushansky ( 2006 ), Lech- ner ( 2007 ), V icente ( 2007 ), R oberts ( 2010 ), S zabolcsi ( 2010 , 2011 ), H artman ( 2011 ), F unak oshi ( 2014 ), Lee ( 2017 ), M atyiku ( 2017 ), H arizano v and Gribano v a ( 2019 ), H arizano v ( 2019 ), Pr eminger ( 2019 ), Landau ( 2020 ), and Sato and M aeda ( 2021 ), among others. The r est of this thesis is dedicated to the pursuit of a unified theory of mo v ement. The empirical 1. See Chapter 2 f or an extensiv e discussion. 4 1.2. A UNITY AP P R O A CH T O M O VEMENT evidence in fa v or f or such an appr oach comes fr om diff er ent cases of v erb disp lacement in Cantonese. The thr ee main claims ar e as f ollo w s, w hich corr esponds to the thr ee main chapters in this thesis. (3) Arguments f or a unified appr oach to mo v ement fr om Cantonese v erb mo v ement a. Chapter 3: H ead mo v ement is constrained b y the same set of locality / minimality r equir e- ments as p hrasal mo v ement ( cf. Chomsky 1995b ; Rizzi 1990 , 2001 , 2004 ); b. Chapter 4: H ead mo v ement exhibits the same range of possible interpr etiv e eff ects as p hrasal mo v ement, and is also constrained b y Scope Econom y (F o x 2000 ); c. Chapter 5: H ead mo v ement chains ar e linearized b y the same mechanism as p hrasal mo v e- ment chains, i.e., C y clic Linearization and cop y deletion (F o x and P esetsky 2005 ). These thr ee arguments f ocus on diff er ent aspects of head mo v ement, namel y , its syntactic pr op- erties (in N arr o w S yntax), its interpr etiv e pr operties (in the syntax -semantic interface ), and the lin- earization of its chain (in the syntax -p honology interface ). They constitute con v erging evidence fr om diff er ent components of the grammar f or the pr oposal that head mo v ement can be tr eated on a par with p hrasal mo v ement. The theor etical consequence of a unified theory of mo v ement is tw o- f old. F irst, it allo w s us to maintain the f ormulation of the structur e-building operation, M erge , in its simp lest f orm. Internal M erge app lies to syntactic constituents without the need to distinguish heads fr om p hrases, in a w a y comparable to External M erge, w hich app lies equall y to both heads and p hrases. Second, it opens up questions of w hether and ho w other r eported diff er ences betw een mo v ement of heads and p hrases can be attributed to components of the grammar other than the mo v ement mechanism. I t should be str essed that the accounts pr oposed f or diff er ent cases of head mo v ement in this thesis does not in v ol v e an y new machinery or princip les of mo v ement. Instead, the crucial ingr edients in these ac - counts ar e independentl y motiv ated b y p hrasal mo v ement, maximizing the exp lanatory po w er of our existing theory of (p hrasal) mo v ement. A f ew r emarks on w hat this thesis is not about ar e in or der . F irst, w hile this thesis f ocuses on head 5 1.3. THE O UTLINE O F THIS THESIS mo v ement, it does not attempt a global alternativ e to v arious cases of head mo v ement. I t does not in v ent new technology or theor etical apparatus specificall y designed f or head mo v ement either . Second, since the thesis f ocuses on the minimal component of mo v ement theories, rather than about a particular appr oach to head mo v ement/ disp lacement in language, some important issues in head mo v ement such as noun/ v erb incorporation and w or d f ormation ar e not discussed. F inall y , w hile this thesis str esses the r ole of the head-p hrase distinction is minimal in f ormulating movement theories , it does not aim at eliminating the primitiv e notions of heads and p hrases in the grammar , w hich r emain imp ortant in the study of, f or examp le, p hrase structur e, pr ojection, labeling algorithm, and so on. 1.3 The outline of this thesis This r est of this thesis is structur ed as f ollo w s. Chapter 2 traces the origin and dev elopment of the notion of head mo v ement since the 1970s. W hile the notion of head mo v ement has pr o v ed empiricall y useful in capturing v arious linguistic p henomena, it also led to debates r elating to theor etical and empirical issues since the earl y minimalist period. I r eview r ecent r esponses to the issues surr ounding head mo v ement. Chapter 3 examines potential interv ening elements in head mo v ement. The discussion builds on f our v erb doubling constructions that come with ( diff er ent ) discourse eff ects. I t is first argued that the v erbs in these constructions undergo mo v ement to the specifier position of a functional head in the left perip hery . I t is further sho wn that, w hile a head does not block the v erb mo v ement, a f ocused element ma y lead to interv ention (i.e., F ocus Interv ention Eff ects ). This pr operty is argued to f ollo w fr om a minimality condition of the operation Agr ee that mak es r ef er ence to syntactic f eatur es (Chomsky 2000 , 2001 ). The findings r ev eal that the H ead M o v ement Constraint does not app l y to all instances of head mo v ement, and that syntactic interv ention eff ects ar e observ ed with head mo v ement, on a par with p hrasal mo v ement. 6 1.3. THE O UTLINE O F THIS THESIS Chapter 4 diagnoses an instance of head mo v ement that induces scope eff ects. I argue that quan- tificational heads such as aspectual v erbs and ( a subset of ) modal v erbs in Cantonese can undergo ( o v ert ) head mo v ement to achiev e scope enrichment. F urthermor e, this mo v ement is constrained b y an econom y condition, Scope Econom y , w hich is independentl y observ ed with mo v ement of p hrasal quantifiers (F o x 2000 ). The findings suggest that head mo v ement is no diff er ent fr om p hrasal mo v e- ment in terms of the potentials to induce semantic eff ects, and that Scope Econom y constrains both head and p hrasal mo v ement. Chapter 5 discusses the issue of ho w mo v ement chains of heads ar e pr onounced and linearized. I t concerns the doubling eff ects of head and p hrase mo v ement in Cantonese. Empirical data r ev eal that the doubling eff ects ar e not specific to mo ving heads and that head mo v ement does not al w a y s lead to doubling eff ects. I t is suggested that doubling eff ects arise fr om the fact that the operation r esponsible f or erasing copies in a mo v ement chain is r egulated b y p honological r equir ements that f ollo w fr om a v ersion of C y clic Linearization (F o x and P esetsky 2005 ). S uch an account deriv es the Cantonese doubling pattern of heads and p hrases without r ecourse to the p hrase structur e status of the ( non-)doubling elements. I t maintains that the mechanism that determines cop y pr onunciation is the same f or head chains and p hrase chains. Chapter 6 concludes the thesis. 7 Chapter 2 A ppr oaching head mo v ement 2.1 Intr oducti on This purpose of this chapter is to trace and r eview the ongoing debates r elating to head moment. In § 2.2 , I discuss the adjunction appr oach to head mo v ement under the Go v ernment and B inding framew ork (Chomsky 1981 , 1986 ). In § 2.3 , I r eview the debates r elating to both the theor etical status and empirical pr operties of head mo v ement under minimalist framew ork (Chomsky 1995b , et seq. ). Then, I discuss tw o major r esponses to the issues r elating to head mo v ement. In § 2.4 , I briefl y o v erview appr oaches that discriminate head mo v ement fr om p hrasal mo v ement ( w hat I r ef er to as non-unity appr oaches ). In § 2.5 , I turn to r ecent pursuits of a unified t heory of mo v ement. 2.2 The origin of head mo v ement 2.2.1 F r om independent tr ansf ormati on rul es to M o v e- I t is w ell observ ed that a v erb ma y appear bey ond its pr ojected v erb p hrase or combine with elements outside the v erb p hrase. T o captur e the deriv ed position of v erbs, earl y pr oposals model v erb dis- p lacement b y positing independent transf ormational rules. Some examp les ar e giv en in ( 4 ). (4) T ransf ormation rules pr oposed to captur e the disp lacement pr operty of v erbs a. Chomsky ( 1957 ): Affix H opping in English 8 2.2. THE O RI GIN O F HEAD M O VEMENT a rule that allo w s tense affix es to be r ealized on the main v erb b. Emonds ( 1970 , 1976 ): have / be - raising in English a rule of “ A UX mo v ement” that r ep laces the modal do with a f ollo wing auxiliary v erb c. Aissen ( 1974 ): V - V mo v ement in causativ e constructions ( e.g., in T urkish ) a rule that “ extracts the embedded V fr om its clause and mo v es it into the matrix clause so that it f orms a v erb unit with the matrix” (p.333) d. Emonds ( 1978 ): V - T mo v ement in F r ench (building on observ ations in Ka yne ( 1975 )) a rule of “F inite V erb Raising” that mo v es the v erb bef or e negation and adv erbs e. den Besten ( 1983 ): Germanic v erb-second p henomenon, F r ench subject -clitic in v ersion and English subject -auxiliary in v ersion a rule of “V erb Pr eposing” that mo v es constituents to the comp lementizer U nder the appr oaches to head movement in K oopman ( 1984 ), T ra vis ( 1984 ), and Bak er ( 1985 , 1988 ), transf ormation rules r elating to v erbs ar e examined under a diff er ent perspectiv e. P articularl y , head mo v ement is characterized as an adjunction rule, allo wing a head X to be (left -)adjoined to another head Y and f orm a comp lex head containing both X and Y . This idea is illustrated in ( 5 ). I r ef er to this characterization as the adjunction appr oach to head movement . (5) H ead mo v ement as head-to- head adjunction ZP Z YP Y X Y XP t X This characterization is primaril y motiv ated b y morp hological considerations w hich allo w a head to “pick up ” additional morp hemes via mo v ement. Empiricall y , this mechanism has pr o v ed useful in 9 2.2. THE O RI GIN O F HEAD M O VEMENT capturing man y diff er ent p henomena, r oughl y classified into tw o main types (f or an o v erview , see R oberts ( 2001 )). (6) a. M orp hological gr o wth of heads (i) noun/ v erb incorporation (ii) v erb mo v ement to T ( e.g. in F r ench ), and to C ( e.g. in v erb-second languages ) b. S urface w or d or der of heads (iii) auxiliary in v ersions of diff er ent sorts in R omance and Germanic languages (iv ) the surface position of v erb in V SO languages ( v ) w or d or der of NP /D P - internal elements in the nominal domain As such, the conception of head mo v ement in ( 5 ) marks an important step to w ar ds unification of mo v ement operations, shifting the discussions fr om construction-specific transf ormation rules to an instance of a mo v ement operation ( adjunction ), and fr om mo v ement of elements of a particular category to mo v ement of elements in a particular structural position. M or e generall y , an imp licit idea behind these appr oaches is ( 7 ), w her e the notion M ove - r epr esents the theor etic pursuit of mo v ement theory that r educes transf ormational rules to a single pr ocess. (7) H ead mo v ement is the case of M o v e- w her e is X 0 . (R oberts 2011 , p.196) In other w or ds, ( 7 ) do not onl y r eflect the attempt to unify diff er ent cases of v erb mo v ement, but also the idea that head mo v ement can be tr eated in a parallel fashion with p hrasal mo v ement. 2.2.2 Constr aints on head mo v ement The suggestion in ( 7 ) is further supported b y the observ ation that head mo v ement as f ormulated in ( 5 ) is constrained b y general w ell - f ormedness conditions that app l y to mo v ement operations and their outputs. I briefl y discuss thr ee of them. F irst, head mo v ement obey s S tructur e Pr eserv ation, w hich is later kno wn as the Chain U nif or - mity Condition. Chomsky ( 1986 ) suggests ( 8 ) as a general condition on mo v ement, w her e mo v ement 10 2.2. THE O RI GIN O F HEAD M O VEMENT operations do not alter the structural status of the mo ving elements (f ollo wing the spirit of S tructur e Pr eserv ation Princip le in Emonds 1970 , 1976 ). (8) a. Onl y a head can be adjoined to a head. b. Onl y a maximal pr ojection can be merged as a specifier . I t is suggested that ( 8 ) f ollo w s fr om mor e other princip les. Chomsky ( 1986 , p.4) suggests that ( 8 a ) “w ould f ollo w fr om an appr opriate f orm of Emonds’ S tructur e- Pr eserving H ypothesis (Emonds ( 1976 )), ” and ( 8 b ) w ould f ollo w “ on the X -bar theor etic assumption that heads cannot be base-generated with- out a maximal pr ojection so that a bar e head cannot appear in the specifier position to r eceiv e a mo v ed X 0 category . ” The f ormulation of head mo v ement in ( 5 ) satisfies S tructur e Pr eserv ation, since a head mo v es to another head position. Second, head mo v ement is said to be subject to locality conditions, with the central one being the H ead M o v ement Constraint (HM C, T ra vis 1984 ). I t r equir es head mo v ement to be app lied in a highl y local w a y . Inf ormall y , it means that head mo v ement cannot skip an interv ening head. A f ormulation of the HM C under the Go v ernment and B inding Theory (Chomsky 1981 , 1986 ) is giv en in ( 9 ). (9) The H ead M o v ement Constraint (T ra vis 1984 , p.131) A head X ma y onl y mo v e into the head Y that pr operl y go v erns X. A c cor dingl y , the step in ( 10 ) is illicit, since Z does not pr operl y go v ern X. Instead, it f or ces head mo v e- ment to be cy clic, as in ( 11 ). 1 (10 ) Illicit cases of head mo v ement ZP X + Z YP Y XP t X 8 (11) C y clic head mo v ement ZP X + Y + Z YP t X + t Y XP t X Thir d, the trace of head mo v ement is subject to the Empty Category Princip le (ECP ), a w ell - 1. F or simp licity , head adjunction is indicated b y the “+” sign betw een heads. 11 2.2. THE O RI GIN O F HEAD M O VEMENT f ormedness condition on traces. The ECP suggests that empty categories such as traces must be pr op- erl y go v erned. (12) The Empty Category Princip le (Lasnik and Saito 1984 , p.240 ) A nonpr onominal empty category mu st be pr operl y go v erned. In eff ect, this r equir es that (head) mo v ement pr oceed in an “up w ar d” fashion, and that XP be the struc - tural comp lement of Y in case of mo v ement of X to Y . A ccor dingl y , head mo v ement as characterized in K oopman ( 1984 ), T ra vis ( 1984 ), and Bak er ( 1985 , 1988 ) is r ecognized as a cor e mo v ement operation, hence an instance of M o v e- . One comp lication, ho w ev er , is that the HM C appears to be specific to head mo v ement, in the sense that p hrasal mo v ement does not exhibit a similar strictl y local natur e. W hile the HM C is motiv ated based on empirical evidence in diff er ent p henomena, it is less clear w h y it shoul d hold in the grammar . In or der to eliminate the particular natur e of the HM C as a specific constraint on head mo v ement, Bak er ( 1988 ) pr oposes to deriv e the HM C eff ects fr om the ECP , w her e he suggests that “ther e must be no barrier category that interv enes betw een the tw o [heads ]” ( original emp hasis, p.55). F or examp le, in the structur e in ( 10 ), Y is “ an interv ening theta assigner [ that ] br eaks a go v ernment path ” (p.56), such that X is not dir ectly theta-connected to its trace, hence does not pr operl y go v erns it. This suggestion highlights the importance of theta-connection in deriving the local natur e of head mo v ement. 2 A diff er ent line to deriv e the HM C suggested b y Rizzi ( 1990 ) mak es r ef er ence to the notion of structural types. H e pr oposes to r elativize minimality eff ects to structural types. R elativized Minimal- ity thus pr o vides an unified account of both the local natur e of head mo v ement and minimality eff ects observ ed in p hrasal mo v ement. A v ersion of R elativized Minimality is giv en in ( 13 ). (13) R elativized Minimality (Rizzi 1990 ; a v ersion tak en fr om Rizzi 2011 , p.221-222) a. In the configuration ... X ... Z ... Y ..., a local r elation cannot connect X and Y if Z interv enes and Z is of the same structural type 2. W hile this captur es the local natur e of cases of incorporation, it appears to be too strict in cases of v erb mo v ement of the R omance/ Germanic kind, w her e, f or examp le, negation and adv erbs ma y interv ene the mo ving head and its trace. 12 2.3. THE D EB A TES SURR O UND IN G HEAD M O VEMENT as X. b. S tructural types: (i) A′ positions, (ii) A positions, and (iii) heads. In eff ect, the structur e in ( 10 ) is disallo w ed because the head Y , being the same structural type as X, “interv enes” betw een X and its trace. As such, the HM C can be subsumed under the general minimality conditions on mo v ement operations. 2.3 The debates surr ounding head mo v ement W hile the adjunction appr oach to head mo v ement ma y be a self -contained notion in the GB-era, the notion of head mo v ement becomes contr o v ersial since the minimalist period. On one hand, head mo v ement does not fit nicel y with the minimalist pursuit started in the 1990s. The particular im- p lementation of head mo v ement ( as adjunction ) violates syntactic princip les that head mo v ement is expected to obey . On the other hand, certain empirical pr operties of head mo v ement ar e often high- lighted because it does not pattern nicel y with p hrasal mo v ement ( such as the locality eff ects and the (lack of ) interpr etiv e eff ects ). I discuss some theor etical concerns in § 2.3.1 , f ollo w ed b y the ( alleged) empirical diff er ences betw een head mo v ement and p hrasal mo v ement in § 2.3.2 . 2.3.1 Theor eti cal concerns of the adjuncti on appr oach to head mo v ement The theor etical concerns f or the adjunction appr oach to head mo v ement ar e w ell - kno wn in the litera- tur e (Chomsky 2001 ; T o y oshima 2000 ; M ahajan 2003 ; T o y oshima 2001 ; S urán yi 2005 ; M atushansky 2006 , among man y others ). I r ep licate the major theor etical concerns under the minimalist assump- tions in Chomsky ( 1995b , 2000 , 2001 ). Bef or e I start, it should be r emark ed that man y of these concerns (if not all) ar e theory - internal. They ma y cease to exist under a diff er ent set of theor etical apparatus and assumptions ( see, e.g., discus- sions in F unak oshi ( 2014 ), chapter 1). H o w ev er , most theor etical princip les or conditions discussed in the f ollo wing subsections ar e independentl y motiv ated b y our understanding of (p hrasal) mo v ement. 13 2.3. THE D EB A TES SURR O UND IN G HEAD M O VEMENT So the primary purpose of this subsection is not to sho w s ho w much the adjunction appr oach deviates fr om a particular theor etical framew ork, but to sho w ho w much the theor etical appearances of head mo v ement diff er fr om that of p hrasal mo v ement. This dir ectl y shapes ho w subsequent pr oposals on head mo v ement r espond to these issues. 2.3.1.1 The Extensi on Conditi on F irst, the adjunction appr oach to head mo v ement appar entl y violates the Extension Condition, w hich is a minimalist r ef ormulation of the S trict C y cle Condition (Chomsky 1973 ). The condition r equir es mo v ement to extend the structur e, or to be eff ected at the r oot node. In ( 5 ), r epeated belo w in ( 14 ), the mo v ement of X does not extend ZP; rather , it is eff ected at Y in a counter -cy clic fashion ( as Y is internal to a pr eviousl y built structur e ). (14) H ead mo v ement as head-to- head adjunction = ( 5 ) ZP Z YP Y X Y XP t X H o w ev er , it should be noted that the v ersion f ormulated in Chomsky ( 1995b ) indeed ex empts head mo v ement ( or mor e general, adjunction ) fr om this condition, as it app lies onl y to substitution. 3 (15) The Extension Condition (Chomsky 1995b , p.190 ) a. [ Generalized T ransf ormation ] and M o v e- extend K to K’ , w hich includes K as a pr oper part. 3. This is in line with his later suggestion that “[h ]ead adjunction ... pr o vides some r eason to w eak en the Extension Condition. ” (Chomsky 2000 , p.137), so as to allo w head adjunction in syntax. H e pr oposes a r elax ed v ersion Extension Condition (i.e. the Least T ampering Condition ), but it ar ouses other issues. F or discussions, see S urán yi ( 2005 ). 14 2.3. THE D EB A TES SURR O UND IN G HEAD M O VEMENT b. S ubstitution operat ions al w a y s extend their target. Another ex empted case is co v ert mo v ement ( e.g. Quantifier Raising), but the co v ert syntactic cy cle is no longer assumed in the subsequent minimalist framew ork (Gr oat and O’N eil 1996 ; P esetsky 1998 ; Chomsky 2000 , 2001 ; Bobaljik 2002 ). This r enders head mo v ement being the onl y ex ception, or one of the v ery f ew ex ceptions to the Extension Condition ( cf. T uck - in mo v ement, N . Richar ds ( 1997 )). M or e specificall y , allo wing head mo v ement to be an ex ception yields tw o kinds of asymmetries in structur e-building ( especiall y , if substitution and adjunction ar e to dissol v e into one single M erge operation ). On one hand, head mo v ement is diff er ent fr om p hrasal mo v ement, w hich typicall y targets the r oot. 4 On the other hand, it diff ers fr om External M erge of a head, w hich al w a y s targets the r oot (i.e., a head cannot be adjoined to another head dir ectl y fr om the N umeration ). This r enders head mo v ement in ( 14 ) a particular case of structur e-building. 2.3.1.2 The Empty Category Princip l e/ Pr oper B inding Princip l e Second, the adjoined head X in ( 14 ) does not c -command its trace t X in a straightf orw ar d w a y , w her e c -command is defined as “X c -commands Y iff the first branching node dominating X dominates Y . ” This suggests that head mo v ement w ould violate the Empty Category Princip le (ECP ) or the Pr oper B inding Condition (P BC). T o ensur e that head mo v ement obey s the ECP or the P BC, it is necessary to r edefine the notion of c -command , f or examp le, b y intr oducing a distinction betw een containment v s. dominance or be- tw een segment v s. category . (16) C-command, based on Bak er ( 1988 , p.36, adapted) X c -commands Y iff X does not dominate Y and f or ev ery maximal pr ojection ZP , if ZP domi - nates X then ZP dominates Y . 4. But see P esetsky ( 2013 ), F unak oshi ( 2012 , 2014 ), and Y uan ( 2017 ) f or pr oposals that some instances of p hrasal mo v e- ment is undermerged / adjoined to another non- r oot p hrasal element. 15 2.3. THE D EB A TES SURR O UND IN G HEAD M O VEMENT (17) C-command, based on Ka yne ( 1994 , p.16) a. X c -commands Y iff X and Y ar e categories and X ex cludes and ev ery category that dominates X also dominates Y ( emp hasis in original). b. X ex cludes Y if no se gment of X dominates Y Either w a y w ould ho w ev er comp licate our definition of c -command. Chomsky ( 2000 ) suggests that they do not “fall under the notion of c -command deriv ed fr om M erge” (p.116), w hich is transitiv e closur e of sisterhood and containment. (18) C-command, v ersion based on Chomsky ( 2000 ), p.116, adapted a. K contains X if K immediat el y contains X or immediatel y contains L that contains X; b. X is a term of K if K contai ns X; c. X c -commands Y if X is the si ster of K that contains Y . 2.3.1.3 The non -successi v e cy cli c natur e Thir d, head mo v ement cannot pr oceed in a successiv e cy clic fashion, as opposed to p hrasal mo v ement ( e.g., successiv e cy clic wh - mo v ement, Chomsky ( 1973 , 1977 )). I t must “pick up ” all the morp hemes in the head position alon g its mo v ement path. In other w or ds, ex corporation is suggested to be impos- sible f or head mo v ement. 5 (19) The Ban on H ead Extraction/ The Ban on Ex corporation ( cf. Bak er 1988 ) If a head X mo v es to Y , then f X + Yg acts as one constituent, i.e., X cannot mo v e out of the head comp lex. 5. But see R oberts ( 1991 ) f or an opposite view . 16 2.3. THE D EB A TES SURR O UND IN G HEAD M O VEMENT (20 ) An illustration of the Ban on H ead Extraction/ the Ban on Ex corporation ZP X + Z YP t X + Y XP t X 8 W hi le such constraint is empiricall y supported b y cases of incorporation and v erb inflection, R oberts ( 2001 ) points out that it does not f ollo w fr om GB conception of head mo v ement. In other w or ds, this ban must be stipulated in the adjunction appr oach to head mo v ement. N ote that it is sug - gested that ( 19 ) ma y be deriv ed fr om the Lexical Integrity H ypothesis, w hich states that morp hological structur e cannot be targeted b y syntactic operation. (21) Lexical Integrity H ypothesis (Lapointe 1980 , p.8) N o syntactic rule can r ef er to elem ents of morp hological structur e. This builds in a morp hological character in head mo v ement, but it then raises an ar chitectural issue concerning the boundary betw een the syntactic component and the morp hological component. 2.3.1.4 Locality constr aints F ourth, as far as locality is concerned, head mo v ement is subje ct to a diff er ent, stricter locality con- dition compar ed to p hrasal mo v ement. The HM C r equir es head mo v ement to be strictl y local, but p hrasal mo v ement need not be so. W hile the HM C eff ects ar e suggested to be subsumed under general conditions on minimality (R elativized Minimality ), the unification with other instances of mo v ement operations ma y be superficial, since the typology of structural types in v ol v es “ at least tw o irr educible distinctions” (Rizzi 2001 , p.90-91). (22) T w o distinctions in structural types a. a distinction betw een heads and p hrases, and in the latter class, b. a distinction betw een positions of arguments (A -positions ) and of non-arguments (A′-positions ). 17 2.3. THE D EB A TES SURR O UND IN G HEAD M O VEMENT In other w or ds, R elativized Minimality still imp licates that the locality constraint on head mo v ement is substantiall y diff er ent fr om p hrasal mo v ement, as heads in general blocks head mo v ement, w hich is not r elativized to another dimension ( as in the cases of p hrasal mo v ement ). 2.3.1.5 The Chain U nif ormity Conditi on F ifth, with the adv ert of Bar e Phrase S tructur e (B PS, Chomsky 1994 ), syntactic pr ojection lev els ar e not exp licitl y assigned to X’ -theor etic categories, and minimal and maximal pr ojections ar e defined in terms of structural r elation, as in ( 23 ). (23) A r elational definition on pr ojection lev els under B PS (Chomsky 1994 ) a. A category that does not pr oject a n y further is a maximal pr ojection XP; b. One that is not a pr ojection at all is a minimal pr ojection X 0 ; c. An y other is an X’ , in visible at the interface and f or computation. This ar ouses a non-trivial concern on the adjunction appr oach to head mo v ement, as it violates the Chain U nif ormity Condition (N unes 1998 ; T o y oshima 2000 , 2001 ). (24) Chain U nif ormity Condition (Chomsky 1995b , p.253) A chain is unif orm with r egar d to p hrase structur e status. ( w her e the p hrase structur e status of an element is its ( r elational) pr operty of being maximal, minimal, or neither ) F or examp le, under ( 23 ), a head (bef or e mo v ement ) is a non- maximal pr ojection ( as it pr ojects ). H o w - ev er , an adjoined head is a maximal pr ojection, as it does not pr oject further . 6 6. T o a v oid this issue, Chomsky ( 1995b ) pr oposes W or d I nterpr etation , pr ocesses that ar e not constrained b y syntactic princip les, but see criticisms in N unes ( 1998 ) and T o y oshima ( 2000 , 2001 ) on the stipulativ e natur e of W or d Interpr etation. 18 2.3. THE D EB A TES SURR O UND IN G HEAD M O VEMENT 2.3.2 Empiri cal diff er ences with p hr asal mo v ement? Apart fr om the theor etical concerns of the adjunction appr oach to head mo v ement, ther e ar e non- trivial contr o v ersies o v er the empirical pr operties of head mo v ement, w hich substantiall y aff ect the w a y head mo v ement is modeled in the grammar . 2.3.2.1 The l ocality constr aints on head mo v ement F irst, w hether the HM C app lies to all instances of head mo v ement is not uncontr o v ersial. I t is ob- serv ed that a substantial amount of cases in v ol v e local head mo v ement, but it is also r eported that head mo v ement can be long -distance, in violation to the HM C. The first type of cases concerns a con- struction w her e a v erb/ particip le is argued to be fr onted bef or e an interv ening auxiliary v erb. This construction is often r ef err ed to as Long H ead M ovement (LHM, Lema and Riv er o 1990 ; Riv er o 1991 ; 1993 ; 1994 ; R oberts 1994 ; W ilder 1994 ; Borsley , Riv er o, and S tep hens 1996 , among others ). (25) Examp les of Long H ead M o v ement a. Bulgarian (Lema and R iv er o 1990 , p.334) Pr očel r ead sum ha v e+pres+1sg knigata book+the ‘I ha v e r ead the book ( comp letel y ). ’ b. Br eton (Borsley , Riv er o, and S tep hens 1996 , p.53) Lennet r ead en 3sg.masc deus has Y ann Y ann al the levr . book ‘ Y ann has r ead the book. ’ Another type of cases comes fr om pr edicate cleft or v erb fr onting constructions. T ypicall y , in these constructions, a v erb is doubled and an extra v erb appears on its o wn in the initial/ perip hery position. 19 2.3. THE D EB A TES SURR O UND IN G HEAD M O VEMENT (26) Examp les of pr edicate cleft/ v erbal fr onting constructions a. V ata (K oopman 1984 , p.159) yī come IJ O s/he w à w ant [ nā N A à w e yī come ] ‘S/he w ants us to come. ’ b. Spanish (V icente 2007 , p.79) Compr ar , buy .INF J uan J ha has dicho said [ que that M aría M ha has compr ado bought un a libr o book ] ‘ As f or buying, J uan has told me that M aría has bought a b ook. ’ I t is an empirical question as to w hether these cases in v ol v e head mo v ement. If they do, then the HM C must be w eak ened. If they do not, the HM C can be r etained as an important pr operty that needs to b e captur ed b y an y pr oposals on head mo v ement. 7 2.3.2.2 The interpr eti v e eff ects of head mo v ement Another diff er ence betw een head and p hrasal mo v ement concerns the interpr etation of mo v ement in the syntax -semantics interface. I t is suggested that head mo v ement is semanticall y inert, as it does not aff ect interpr etation. F or examp le, ther e is no scope or r econstruction eff ects in case of head mo v e- ment (Chomsky 2001 ; H arley 2004 ; Platzack 2013 ). This is especiall y true f or v erb mo v ement cases in the inflectional sy stem in the sense of Germanic/R omance languages ( see chapter 4 f or mor e discus- sions ). Chomsky ( 2001 ) points out that “the semantic eff ects of head- raising in the cor e inflectional sy stem ar e slight or nonexistent” (p.37). If head mo v ement in general does not aff ect interpr etation, this w ould r epr esent a substantial diff er ence fr om p hrasal mo v ement to the extent that the mecha- nism held r esponsible f or p hrasal mo v ement might be considerabl y diff er ent fr om the one f or head mo v ement. H o w ev er , a number of cases of head mo v ement ha v e been r eported to be able to bring along se- 7. F or discussions of the locality constraints on head mo v ement, see Chapter 3. 20 2.3. THE D EB A TES SURR O UND IN G HEAD M O VEMENT mantic eff ects. T able 2.1 summarizes some r ecent pr oposals in support of this view . H ead Language Scope eff ects R ef er ence( s ) Determiner J apanese enhanced r estriction T akahashi ( 2002 ) N egation English NP I licensing R oberts ( 2010 ) and S zabolcsi ( 2010 ) J apanese NP I licensing Kishimoto ( 2007 ) English v arieties outscope subjects M atyiku ( 2017 ) and Landau ( 2020 ) K or ean outscope objects H an, Lidz, and M usolino ( 2007 ) J apanese outscope objects Sato and M aeda ( 2021 ) M odal v erb English outscope subjects Lechner ( 2007 , 2017 ) English outscope negation I atridou and Zeijlstra ( 2013 ) and H omer ( 2015 ) Aspectual v erb Shupamem outscope subjects S zabolcsi ( 2010 , 2011 ) T able 2.1 : S ummary of evidence of scope eff ects w ith head mo v ement The anal y ses in these pr oposals ar e not uncontr o v ersial. Debate continues as to w hether head mo v ement is indeed in v ol v ed in these cases. So w hether the lack of semantic eff ects is an inher ent pr operty of head mo v ement is an unsettled empirical question. The answ er to this question aff ects ho w head mo v ement should be modeled in mo v ement theories. 2.3.2.3 The morp ho-p hono l ogi cal r ealizati on of he ad mo v ement As opposed to p hrasal mo v ement, ther e is an important morp hological character in head mo v ement. F or examp le, head mo v ement is usuall y motiv ated b y w or d f ormation or affixation, as in cases of incorporation and v erb inflection. 8 H o w ev er , it is also suggested that such morp hological character ma y not be unique to head mo v ement. Some instances of p hrasal mo v ement ma y lead to w or d f orma- tion, such as the Sax on genitiv e mark er ’ s (Giorgi and Longobar di 1991 ). Also, the so-called “ sno wball mo v ement” in F innish in v ol v e p hrases “r olling up ” along the mo v ement path (H uhmarniemi 2012 ), in a w a y similar to head mo v ement. These cases suggest that the morp hological character ma y not be unique to head mo v ement. 8. But this is not necessaril y so, since T -to-C mo v ement in man y cases does not lead to incr eased morp hological com- p lexity . See also discussions in H arizano v and Gribano v a ( 2019 ). 21 2.4. N O N -UNITY AP P R O A CHES T O HEAD AND P HRASAL M O VEMENT In terms of p honological r ealization, especiall y in the discussions of pr edicate clefts, head mo v e- ment commonl y leads to double pr onunciation of the mo v ement chains. This is in contrast with p hrasal mo v ement, w her e double pr onunciation is far less common ( though not unattested). On one account, a doubled head is suggested to be a consequence of morphological fusion with another head (N unes 1995 , 2004 ), such that the tw o members in a head chain ha v e to be pr onounced as they ar e dis- tinct syntactic objects (pr o vided the cop y deletion operation onl y targets identical objects ). W hether these issues should be tak en into consideration in a theory of head mo v ement is a non-trivial issue. 2.4 N on -unity appr oaches to head and p hr asal mo v ement Ther e ar e tw o major r esponses to the issues surr ounding head mo v ement, to be discussed in this sec - tion (§ 2.4 ) and the next (§ 2.5 ). The first type of r esponses ackno w ledges the theor etical and empirical diff er ences betw een head mo v ement and p hrasal mo v ement, and tr eat head mo v ement diff er ent fr om p hrasal mo v ement. I call appr oaches along this line non-unity appr oaches . Either head mo v ement is eliminated fr om the syntax or head mo v ement is radicall y r ef ormulated. In light of r ecent compr ehensiv e assessments and discussions of the diff er ent appr oaches to head mo v ement, as in R oberts ( 2011 ) and Dékán y ( 2018 ), the discussion in this subsection is intentionall y brief, with the aim of sho w casing the major dir ections of the dev elopment of head mo v ement in the literatur e. 2.4.1 Eliminating head mo v ement fr om the s yntax 2.4.1.1 P ost -s yntacti c mo v ement/ oper ati ons S ince the theor etical concerns of head mo v ement arise due to violations of syntactic princip les, a pos- sible solution is that head mo v ement is r esulted fr om operations that ar e not in narr o w syntax, but in the post -syntactic component (i.e., in the P F branch of grammar ). 22 2.4. N O N -UNITY AP P R O A CHES T O HEAD AND P HRASAL M O VEMENT Ther e ar e at least tw o w a y s to imp lement this idea. The first one is to posit some post -syntactic mo v ement operation in the P F branch (Chomsky 2001 ; Boeckx and S tjepano vić 2001 ; Schoorlem- mer and T emmerman 2012 ; H arizano v and Gribano v a 2019 ). 9 F or examp le, the imp lementation in H arizano v and Gribano v a ( 2019 ), r esorts to post -syntactic head Raising/Lo w ering ( cf. Embick and N o y e r 2001 ). Essentiall y , it transp lants the adjunction appr oach to head mo v ement to the P F branch. Another w a y of imp lementation denies the existence of mo v ement in the P F branch. Instead, the disp lacement pr operty of a head is r esulted fr om the w a y syntactic structur e is linearized, i.e., the linearization instructions to p honology (B r ody 2000 ; A dger 2013 ; Platzack 2013 ; Ramchand and S v enonius 2014 ; H all 2015 ; S v enonius 2016 ). This w ould in v ol v e a substantiall y diff er ent mechanism to captur e disp lacement pr operty , w hen compar ed to p hrasal mo v ement ( see Dékán y 2018 , f or dis- cussions ). 2.4.1.2 R emnant p hr asal mo v ement A diff er ent line of appr oaches seeks to r eanal yze head mo v ement as r emnant p hrasal mo v ement, such that a head does not mo v e on its o wn, but a p hrase containing onl y a head does. In other w or ds, w hat is said to be head mo v ement is actuall y p hrasal mo v ement in disguise. These appr oaches maintain that the mechanism r esponsible f or head mo v ement/ disp lacement is syntactic, but ther e is no syntactic head mo v ement at all, since the head does not mo v e. This idea is adopted b y K oopman and S zabolcsi ( 2000 ), M assam ( 2000 ), Rack o w ski and T ra vis ( 2000 ), M ahajan ( 2003 ), N ilsen ( 2003 ), and M üller ( 2004 ), among man y others. In terms of imp lementation, a r emnant p hrasal mo v ement appr oach head mo v ement r elies hea vil y on the mechanism that ev acuates the el - ements within a p hrase (but stranding a head). I t is onl y after other p hrase- internal elements ha v e mo v ed out that the p hrase mo v es as if the head mo v es on its o wn. 9. H arizano v and Gribano v a ( 2019 ) do not den y the pr esence of syntactic head mo v ement, ho w ev er . 23 2.4. N O N -UNITY AP P R O A CHES T O HEAD AND P HRASAL M O VEMENT 2.4.2 R ef ormulating head mo v ement in the s yntax Other pr oposals seek to r ef ormulate the syntactic imp lementation of head mo v ement such that it obey s the syntactic princip les that mo v ement operations ar e expected to obey . In other w or ds, in these appr oaches, heads do mo v e in the syntax, but they mo v e in a w a y diff er ent fr om p hrases. F or examp le, N unes ( 1995 , 2004 ), Bobaljik and B r o wn ( 1997 ), and U riager eka ( 1998 ) pr opose a sidewar d movement appr oach to head mo v ement, w her e mo v ement occurs acr oss tw o parallel w orkspaces. This specificall y a v oids the violation to the Extension Condition (but other concerns r emain, e.g., the violation to the ECP and the CU C). Another imp lementation is suggested in K oeneman ( 2000 ), Bury ( 2003 ), F anselo w ( 2003 ), S urán yi ( 2005 , 2008 ), and Donati ( 2006 ), w ho pr opose that head mo v ement in v ol v es r epr ojective movement . The cor e idea is that a head mo v es into another empty head position, and pr ojects a second time ther e. This simultaneousl y a v oids violations to the Extension Condition, the CU C and the ECP . Y et another type of appr oaches pr opose a syntactic operation/ mo v ement that allo w s the disp lace- ment of the p honological f eatur es of a head, such as Conflation (H arley 2004 ; cf. H ale and K ey ser 2002 ), and ( syntactic ) p honological mo v ement in the sense of Z w art ( 2001 ). A similar but not identical oper - ation, Gener alized H ead M ovement , is r ecentl y pr oposed in Arr egi and Pietraszk o ( 2021 ), allo wing the specific parts of a single comp lex head to be associated with diff er ent terminal nodes. On the other hand, R oberts ( 2010 ) pr oposes that head mo v ement is achiev ed via A gr ee , f ollo w ed b y an incorporation(- lik e ) operation. This appr oach is also adopted in A elbr echt and Dikk en ( 2013 ) and I orio ( 2015 ). The idea o f this appr oach is that, b y Agr ee- ing with a lo w er head, a higher head acquir es all the f ormal f eatur es of the lo w er head. B y virtue of this, the lo w er head becomes def ectiv e, and is subsequentl y incorporated into the higher head. 24 2.5. RECENT PURSUITS O F A UNIFIED THEO R Y O F M O VEMENT 2.4.3 Interim summary I t should be r emark ed that the v alidity and legitimacy of these appr oaches ar e both an empirical ques- tion and a theor etical question. I t might be that one of these appr oaches turns out to be a global alternativ e to head mo v ement, or that these diff er ent appr oaches ar e all needed f or diff er ent head disp lacement/ mo v ement p henomena. E v aluation of these appr oaches is bey ond the scope of this the- sis. The r elev ance of these appr oaches to this thesis, ho w ev er , is that a conceptual question f or the non-unify appr oaches r emains: w hat pr ev ents head mo v ement in syntax, or mor e specificall y , w hat pr ev ents a head, being a constituent, fr om undergoing mo v ement in a w a y similar to p hrases. Put diff er entl y , it begs the question of ho w the diff er ences betw een head and p hrasal mo v ement f ollo w fr om general princip les of the grammar . 2.5 R ecent pur suits of a unifi ed theory of mo v ement The second type of r esponses to the issues r elating to head mo v ement is to assimilate head mo v ement to p hrasal mo v ement as much as possible, maintaining that head mo v ement is an instance of M o v e- or is achiev ed b y Internal M erge (Chomsky 2001 ). The general idea is that if head mo v ement is pr oblematic because it in v ol v es a mechanism that deviates fr om the one f or p hrasal mo v ement, this can be a v oided if head mo v ement is achiev ed via the same mechanism. The pursuit of a unified theory of mo v ement is r eflected on r ecent r e-ev aluations of the diff er ences betw een head mo v ement and p hrasal mo v ement (T o y oshima 2001 ; M atushansky 2006 ; Lechner 2007 ; V icente 2007 ; H artman 2011 ; F unak oshi 2014 ; H arizano v and Gribano v a 2019 ; H arizano v 2019 ; Pr eminger 2019 , among others ). In w hat f ollo w s, I briefl y discuss some of them. 25 2.5. RECENT PURSUITS O F A UNIFIED THEO R Y O F M O VEMENT 2.5.1 H ead mo v ement to the specifi er positi on One straightf orw ar d w a y to a v oid the theor etical concerns of head mo v ement is to allo w a head to mo v e into the specifier position, in a w a y similar to p hrasal mo v ement. This possibility has its r oot in K oopman ( 1984 ), w her e a v erb is suggested to be ablle to mo v e into a V’ -position ( cf. A ’ -position ). This is further discussed in F ukui and T akano ( 1998 ), T o y oshima ( 2000 , 2001 ), M atushansky ( 2006 ), and V icente ( 2007 ). R ecall that the theor etical space f or f ormulating head mo v ement is sev er el y r estricted in the GB- era. A head-to- head adjunction appr oach is almost the onl y possible f ormulation at that time 10 . Other imp lementations of head mo v ement, such as head mo v ement to a specifier position or head mo v ement adjoining to a p hrase, ar e ruled out due to the combined eff ects of the X’ -theor etic assumption of p hrase structur e and the assumption of S tructur e Pr eserv ation (Emonds 1970 , 1976 ; Chomsky 1986 ). In other w or ds, mo v ement operations ar e expected to obey the CU C. H o w ev er , with the B PS r ep lacing the X’ -theor etic p hrase structur e, the CU C ev en rules out the adjunction appr oach to head mo v ement. I t is suggested that the CU C should be abandoned fr om the grammar (N unes 1998 ; T o y oshima 2000 , 2001 ). This w ould not onl y allo w (i) head-to- head adjunc - tion, but also (ii) head-to-specifier mo v ement and (iii) head adjoining to a p hrases. 11 On the other hand, F ukui and T akano ( 1998 ) suggest that the CU C can be maintained, if “uni - f ormity” is based on non-distinctness (instead of identity ). If so, head mo v ement does not necessaril y violate the CU C. This is because w hile the head in the launching position and the head in the landing position heads ar e not identical in terms of pr ojection lev el, they ar e not distinct , as both of them ar e minimal pr ojections (i.e., they ar e not pr ojected). Lik ewise, this w ould not onl y allo w head-to- head adjunction, but also head-to-specifier mo v ement and head adjunction to a p hrase. 12 10. H ead mo v ement ma y be f ormulated as a substitution rule, as pr oposed in Rizzi and R oberts ( 1989 ). 11. Abandoning the CU C w ould also allo w a p hrase to pr oject again after mo v ement ( w hich is pr eviousl y ruled out b y the CU C), but N unes ( 1998 ) suggests that this can be ruled out independentl y . 12. This r evision of the CU C w ould still rule out further pr ojection of a p hrase after mo v ement ( as intended b y the CU C), since the p hrase ( r e-)pr ojecting in the landing site w ould be non- maximal and non- minimal, hence in visible, under the B PS. I t is thus distinct fr om the p hrase in the launching site. 26 2.5. RECENT PURSUITS O F A UNIFIED THEO R Y O F M O VEMENT Either the abandonment or the r evision of the CU C opens up new possibility to f ormulate head mo v ement within the minimalist framew ork. The possibility of head-to-specifier mo v ement is adopted in man y subsequent w orks ( e.g., M atushansky 2006 ; V icente 2007 ; Cheng and V icente 2013 ; H arizano v and Gribano v a 2019 ; H arizano v 2019 ; B. H su 2021 , to name just a f ew ). I t should be noted that head mo v ement to the specifier position is often tak en as the first step of deriving pr operties of head mo v ement, w hich ma y be f ollo w ed b y some morp hological operation as in M atushansky ( 2006 ) and H arizano v and Gribano v a ( 2019 ), or another syntactic operation, e.g., Coalescence as in B. H su ( 2021 ). 2.5.2 N o head -specifi c l ocality constr aint W ith r egar d to the particular locality constraint of head mo v ement (i.e., the HM C), a gr o wing body of evidence suggests that the local natur e of head mo v ement is not an inher ent pr operty of head mo v e- ment in general, as briefl y discussed in § 2.3.2.1 . Specificall y , the rich literatur e of pr edicate cleft acr oss languages sho w that man y instances of head mo v ement does not obey the HM C. M or e importantl y , head these instances of head mo v ement is constrained in a w a y similar to p hrasal mo v ement, w her e they exhibit sensitivity to syntactic islands, w hile tolerating clausal boundaries, mo ving in a long - distance fashion ( e.g., V icente 2007 ; H ein 2018 ; H arizano v and Gribano v a 2019 ; Pr eminger 2019 , among man y others ). On the other hand, the local natur e of head mo v ement ma y be due to C( ategory )-selection, as suggested in M atushansky ( 2006 ). S ince C-selection is b y definition local, the dependency betw een a head and the head that it C-selects must be local. N ote that C-selection is also local betw een a head and its selecting comp lement. A ccor dingl y , the local natur e of head mo v ement has nothing to do with the inher ent pr operty of mo v ement operations. N o special locality constraint need to be posited f or head mo v ement. 13 13. The locality issues r elating to head mo v ement will be discussed in gr eater details in Chapter 3. 27 2.6. SUMMAR Y 2.5.3 The interpr etati on of head mo v ement W hile the lack of interpr etativ e eff ects of head mo v ement is often tak en to motiv ate a non-syntactic appr oach, it does not necessaril y rule out the syntactic natur e of head mo v ement. In other w or ds, w hile the interpr etativ e eff ects of head mo v ement ar e evidence that head mo v ement must r eside in the syntactic component, their absence does not speak against the syntactic natur e of head mo v ement. Certain instances of p hrasal mo v ements ma y also lack interpr etiv e eff ects f or diff er ent r easons. M a- tushansky ( 2006 ) also suggests that the r eason w h y v erb mo v ement usuall y lack interpr etiv e eff ects is due to their semantic type, w hich does not aff ect interpr etation no matter a v erb is interpr eted in the launching position or the landing position. Pr oposals that argue f or the interpr etiv e eff ects of head mo v ement, as mentioned in § 2.3.2.2 , ar e thus potential evidence f or a unity appr oach to head and p hrasal mo v ement, w her e both of them ar e syntactic b y natur e. 14 2.6 S ummary The ongoing debates of head mo v ement imp lies that head mo v ement/ disp lacement ma y be a non- unif orm p henomenon. The div erse natur e ma y be due to the fact that the disp lacement pr operties of heads can be attributed to diff er ent operations in diff er ent components of the grammar . H o w ev er , I str ess that this does not necessaril y r eflect the non-unif orm natur e of movement oper ations in syntax. The r est of the thesis is dedicated to a minimal pursuit of the question of to w hat extent a unified theory of mo v ement, specificall y one that does not distinguish heads fr om p hrases, is possible. I pr esent evidence fr om Cantonese v erb mo v ement sho wing that at least some instances of head mo v ement beha v e exactl y the same as p hrasal mo v ement. 14. I t should be ackno w ledged that the arguments pr esent f or the interpr etiv e eff ects ar e not uncontr o v ersial. See Chap- ter 4 f or extensiv e discussions. 28 Chapter 3 Interv enti on eff ects: v erb mo v ement to pe- rip her al positi ons Chapter S ummary: This chapter ex amines potential intervening elements in head movement by investigating f our cases o f non-local verb displacement in Cantonese. I n these cases, the verbs ar e doubled, and their copy appears in the initial or final position o f the sentence. I pr opose that these f our cases unif ormly involve head movement to a specifier position in the CP periphery, in a way identical to their phr asal counterparts. I further argue that elements o f the same structur al types ( i.e., heads/ verbs ) do not necessarily block the pr oposed movement; instead, elements that possess the same syntactic f eatur e ar e genuine interveners. The findings in the chapter chal- lenge the status o f the H ead M ovement Constr aint as a gener al constr aint on head movement. A t the same time, I show that the pr oposed head movement exhibits the syntactic interven- tion eff ects that ar e observed with phr asal movement. I conclude that head movement is not constr ained in a way diff er ent fr om phr asal movement with r egar d to intervention. P articu- larly, intervention eff ects ar e calculated in terms o f syntactic f eatur es but not structur al types. This conclusion necessitates a movement theory that does not distinguish head movement fr om phr asal movement in terms o f locality. 29 3.1. INTR O D U CTI O N 3.1 Intr oducti on This chapter examines potential interv ening elements in head mo v ement b y in v estigating f our cases of non- local v erb disp lacement in Cantonese. I t is often suggested that head mo v ement is subject to stricter locality r equir ements than p hrasal mo v ement, in the sense that an interv ening head w ould block head mo v ement. This is commonl y kno wn as the H ead M o v ement Constraint (T ra vis 1984 ; Bak er 1988 ; Rizzi 1990 , among others ). H o w ev er , I will sho w that a head is not necessaril y an inter - v ener in head mo v ement. F r om the perspectiv e of a head mo v ement anal y sis, the cases of non- local v erb disp lacement in Cantonese sho w that head mo v ement is interv ened b y elements possessing the same syntactic f eatur e, but not elements of the same structural types (i.e., heads ). 1 I start with a brief intr oduction of the r elev ant constructions in Cantonese. The canonical w or d or der in Cantonese is S( ubject )- V( erb )-O( object ), and sentence particles typicall y appear at the end of the sentence. H o w ev er , non-canonical w or d or der is commonl y f ound, and it is emp lo y ed to con v ey diff er ent inf ormation structural meanings. Examp les include (i) topic constructions (M atthew s and Y ip 2011 ; f or M andarin, Chao 1968 ; Li and Thompson 1981 , i.a.), (ii) ‘ even’- f ocus constructions (f or M an- darin, P aris 1979 , 1998 ; Sh yu 1995 ; Badan 2007 ; Constant and Gu 2010 ); (iii) copula f ocus constructions (C. C.-H. Cheung 2015 ; f or M andarin, L. L. S. Cheng 2008 ; P an 2014 , 2017 , 2019 ) 2 ; and (iv ) right dis - location (L. Y .-L. Cheung 1997 , 2005 , 2009 ; La w 2003 ; B. H.-S. Chan 2013 ; Lee 2017 , 2020 ; Lai 2019 ; K.-F . Y ip 2020 ). An examp le of each construction is giv en in ( 27 ), r espectiv el y . F or illustrativ e pur - poses, objects ar e chosen to demonstrate the change in w or d or der , but these constructions ar e b y no means ex clusiv e to objects. 1. Thr oughout this chapter , I use the term structur al types to r ef er to heads and phr ases . 2. In V ictor P an ’ s w ork, these constructions ar e r ef err ed to as ex -situ cleft constructions . I a v oided this name because it does not distinguish them fr om ‘ ev en ’ - f ocus constructions and it pr esumes a cleft structur e. S ince these constructions obligatoril y in v ol v e a copula bef or e the f ocus, I adopt a mor e descriptiv e name. 30 3.1. INTR O D U CTI O N (27) N on-canonical w or d or der with r egar d to objects a. O S V : T opic constructions (M atthew s and Y ip 2011 , p.84) Zoeng cl zi paper nei y ou f ong put hai at bin w her e aa3? sfp ‘W her e do/ did y ou put the paper?’ b. Lin - O S V : ‘E v en ’ - f ocus constructions ( cf. Sh yu 1995 , p.6) Lin ev en ni -bun this-cl s yu book Aaming Aaming dou also maai - zo. buy -perf ‘ Aaming ev en bought this book. ’ c. H ai - O S V : Copula f ocus constructions ( cf. P an 2014 , p.19) H ai co p nei y ou ge m o d tai dou attitude k eoidei they m- zungji not - lik e ze1. sfp ‘I t is onl y y our attitude that they do not lik e. ’ d. S V sfp O : Right dislocation (L. Y .-L. Cheung 1997 , p.1 2, adapted) Daaigaa w e dou all m- zi not - kno w lo1 sfp ni - joeng this-cl je . thing ‘ All of us don ’t kno w this thing. ’ This chapter f ocuses on the v ariants of these f our constructions in Cantonese, all of w hich in v ol v e a doubled v erb in a non-canonical position. An examp le of each construction intr oduced in ( 27 ) is giv en in ( 28 ). 31 3.1. INTR O D U CTI O N (28) N on-canonical w or d or der with r egar d to v erbs a. V S V O: T opic const ructions of v erbs ( cf. Cheng and V icente 2013 , p.13) M aai buy k eoi s/he hai co p maai -gw o buy -exp go-bun that -cl syu. book ‘ As f or buying, s/he has bought that book (but. ..). ’ b. Lin - V S V O: ‘E v en ’ - f ocus constructions of v erbs ( cf. Cheng and V icente 2013 , p.2) Lin ev en tai r ead k eoi s/he dou also m- tai not - r ead ni -bun this-cl syu. book ‘S/he didn ’t ev en READ this book. ’ c. H ai - V S V O: Copula f ocus constructions of v erbs H ai co p dim touch Aaming Aaming m-gam not -dar e dim touch ni - zek this-cl dungmat animal ze1. sfp ‘ Aaming dar e not to T O U CH this animal onl y . ’ d. S V O sfp V : Right dislocation/ dislocation cop ying of v erbs (K. K. Chan 2016 , p .18, adapated) Zoengsaam Zoengsaam gammaan tonight fan sleep ni - zoeng this-cl cong bed aa3 sfp fan . sleep ‘Zoengsaam ( will) sleep on this bed toni ght. ’ In these cases, a cop y of the main v erb appears in the left or right perip hery of the sentence. All these sentenecs in ( 28 ) with the disap lced v erb ar e associated with diff er ent discourse eff ects. These v ariants ha v e r eceiv ed limited attention in the Cantonese literatur e. Cases of v erbs appearing in topic constructions as in ( 28 a ) and ‘ ev en ’ - f ocus constructions as in ( 28 b ) ar e discussed in M atthew s and Y ip ( 1998 , 2011 ). 3 T o the best of m y kno w ledge, copula f ocus constructions of v erbs as in ( 28 c ) ar e not discussed in the literatur e. The cases of right dislocation as in ( 28 d) ar e sometimes r ef err ed to as 3. Although the M andarin counterparts of the sentences in ( 28 ) r eceiv e some attention, the discussion is still v ery limited. As far as I am a w ar e, P aris ( 1998 ) and Liu ( 2004 ) f ocus on the inf ormation structural status of the doubled v erbs, w her eas Constant and Gu ( 2010 ), Cheng and V icente ( 2013 ), and Y ang and W u ( 2019 ) off er some lev el of anal ytical anal y sis. 32 3.1. INTR O D U CTI O N dislocation copying , since the dislocated element is not associated with a “ gap ” in the original sentence, as opposed to or dinary right dislocation. Dislocation cop ying has r eceiv ed r elativ el y mor e attention in the literatur e, as discussed in L. Y .-L. Cheung ( 2015 ), T ang ( 2015 ), K. K. Chan ( 2016 ), and Lai ( 2019 ). The w or d or der patterns of S, V and O illustrated in ( 27 ) and ( 28 ) can be summarized in T able 3.1 . F or con v enience, I r ef er to these constructions targeting v erbs as verb dou bling constructions . 4 (i) topic (ii) ‘ ev en ’ - f ocus (iii) copula f ocus (iv ) right dislocation Object O S V lin - O S V hai - O S V S V sfp O V erb V S V O lin - V S V O hai - V S V O S V O sfp V T able 3.1 : The w or d or der patterns illustrated in ( 27 ) and ( 28 ) Disr egar ding the doubling eff ects, w hich I will set aside thr oughout this chapter , T able 3.1 sho w s that p hrasal dislocation of the object and v erb dislocation pattern identicall y . 5 After a description of the syntactic and semantic pr operties of these v erb doubling constructions, I argue f or tw o claims. F irst, I pr opose that they unif orml y in v ol v e head mo v ement to a specifier position in the CP perip hery , in a w a y similar to their p hrasal counterparts. 6 I justify a mo v ement anal y sis with evidence fr om (i) lexical identity eff ects; (ii) island eff ects, and (iii) idiomatic expr essions. Second, I argue that the pr oposed mo v ement exhibit interv ention eff ects that ar e calculated based on syntactic f eatur es, but not structural types. The evidence comes fr om the observ ation that the tw o v erbs in v erb doubling constructions do not tolerate an interv ening f ocused element , but they allo w an interv ening head . The findings in the chapter challenge the status of the H ead M o v ement Constraint (T ra vis 1984 ) as a general constraint on head mo v ement. Importantl y , it r ev eals that head mo v ement exhibits f ocus interv ention eff ects that ar e also observ ed with p hrasal mo v ement (Rizzi 1990 , 2001 , 2004 ). I con- 4. Ther e ar e man y other instances of v erb doubling in Cantonese, but they do not necessaril y giv e rise to discourse eff ects or do so in a similar w a y . I confine the discussion to these f our cases. 5. The diff er ence in doubling eff ects betw een objects and v erbs is non-trivial and deserv es an exp lanation. Chapter 5 is dedicated to this issue. 6. This is not to sa y that all instances of constructions with a disp laced object necessaril y in v ol v e mo v ement. The surface object ma y be base generated ther e and associated with an empty category in the canonical object position (f or examp le, see, H uang, Li, and Li 2009 ; Sh yu 1995 ; C. C.-H. Cheung 2008 , 2015 ). 33 3.2. INTER VENTI O N EFFECTS AND HEAD M O VEMENT clude that in terms of interv ention locality , head mo v ement is not constrained in a w a y diff er ent fr om p hrasal mo v ement. These necessitate a mo v ement theory that does not distinguish head mo v ement fr om p hrasal mo v ement in terms of locality . This r est of this chapter is organized as f ollo w s. § 3.2 r eview s the discussions on the interv ention eff ects on head mo v ement. § 3.3 off ers a detailed description on the f our v erb doubling constructions. § 3.4 pr esents evidence f or a v erb mo v ement anal y sis. § 3.5 examines the potential interv eners in v erb doubling constructions. § 3.6 details the pr oposal. § 3.7 discusses tw o families of alternativ e anal y ses to a head mo v ement appr oach. § 3.8 exp lor es some consequences of the pr oposal. § 3.9 concludes the chapter . 3.2 Interv enti on eff ects and head mo v ement This section r eview s intervention eff ects in head mo v ement. I adopt a w orking definition of interv en- tion suggested in Rizzi ( 2011 )). (29) Interv ention (Rizzi 2011 , p.220 ) A local r elation is disrupted b y the interv ention of an element with certain qualities w hich mak e it a potential participant in the local r elation. One pr ominent quality of interv eners in the discussion of head mo v ement is one that concerns the structural type of an element. Specificall y , mo v ement dependencies betw een heads ar e said to be disrupted b y interv ening heads, commonl y kno wn as the H ead M o v ement Constraint (hencef orth HM C, T ra vis 1984 ). I discuss the empirical and conceptual motiv ations of HM C in § 3.2.1 . Then, I turn to its particular natur e and potential ex ceptions to the HM C in 3.2.2 . I further discuss pr oposals in def ense of HM C in § 3.2.3 . 34 3.2. INTER VENTI O N EFFECTS AND HEAD M O VEMENT 3.2.1 Interv enti on due to i denti cal structur al types H ead mo v ement is f ormulated in K oopman ( 1984 ), T ra vis ( 1984 ), and Bak er ( 1988 ) as a syntactic op- eration that in v ol v es head-to- head adjunction, w her e a lo w er head mo v es up and adjoins to a higher head, as in ( 30 a ). A head ma y be disp laced o v er a long distance as long as the head mo v e thr ough all the interv ening head positions ( and pick up the heads ), as in ( 30 b ). (30 ) H ead mo v ement as head-to- head adjunction a. ZP Z YP X + Y XP t X b. ZP X + Y + Z YP t X + t Y XP t X I t is also suggested that head mo v ement is subject to locality constraints that ar e stricter than p hrasal mo v ement. F or examp le, head mo v ement is said to be subject to the H ead M o v ement Constraint (HM C, T ra vis 1984 ), w hich r equir es head mo v ement to be highl y local. A f ormulation of the HM C under the Go v ernment and B inding Theory (Chomsky 1981 , 1986 ) is giv en in ( 31 ). (31) The H ead M o v ement Constraint, in T ra vis ( 1984 , p.131) A head x ma y onl y mo v e into the head y that pr operl y go v erns x. S ince the notion of pr oper go v ernment has been abandoned in the modern minimalist models (Chom- sky 1995b , et seq. ), I adopt a v ersion of the HM C r ef ormulated in R oberts ( 2001 ). This v ersion still captur es the original insight of the HM C. (32) The H ead M o v ement Constraint, in R oberts ( 2001 , p.113) H ead mo v ement of X to Y cannot skip an “interv ening” head Z. ( w her e Z interv enes betw een Y and X iff Y asymmetricall y c -commands both X and Z, w hile Z asymmetricall y c -commands X) Configurationall y , cases that violate the HM C ar e illustrated in ( 33 ). In both cases, Y is “ skipped” w hen 35 3.2. INTER VENTI O N EFFECTS AND HEAD M O VEMENT X mo v es to Z. 7 (33) Cases of head mo v ement that violates of the H ead M o v ement Constraint a. ZP X + Z YP Y XP t X 8 b. ZP X + Z YP t X + Y XP t X 8 The empirical evidence of the HM C comes fr om v erb/ auxiliary mo v ement in Germanic and R o- mance languages. F or examp le, in English, an auxiliary can mo v e to C as long as ther e is no interv ening auxiliary . Thus, onl y the mo v ement in ( 34 b ) is allo w ed, as opposed to ( 34 b ). (34) English auxiliary mo v ement (Rizzi 1990 , p.11) a. They could ha v e left. b. Coul d they < could> ha v e left? c. * H a v e they could <ha v e> left? head movement skipping a T position S imilarl y , in I talian, w hile both auxiliaries and particip les can mo v e to C in certain non- finite clause, as in ( 35 a-b ), the particip le cannot mo v e acr oss the auxiliary , sho wn in ( 35 c ). (35) I talian v erb mo v ement to C (Rizzi 2001 , p.93) a. Essendo M ario <essendo> tornato a Milano, . . . “H a ving M ario come back to Milan, . . . ” b. T ornato M ario <tornato> a Milano, . . . “Come back M ario to Milan, . . . ” 7. The w a y HM C as stated in ( 32 ) also pr ev ents a head fr om mo ving out of a head comp lex, w hich is often r ef err ed to as an independent constraint on head mo v ement, as giv en in (i). (i) The Ban on H ead Extraction/ The Ban on Ex corporation ( cf. Bak er 1988 ) If a head X mo v es to Y , then f X + Yg acts as one constituent, i.e., X cannot mo v e out of the head comp lex. 36 3.2. INTER VENTI O N EFFECTS AND HEAD M O VEMENT c. * T ornato M ario essendo <tornato> a Milano . . . participle m ovement skipping the auxiliary “Come back M ario ha ving to Milan, . . . ” The HM C is subsumed under R elativized Minimality (Rizzi 1990 ), w hich attempts to unify similar interv ention eff ects observ ed with A - mo v ement and A ’ - mo v ement. R elativized Minimality specifi - call y mak es r ef er ence to structural types, w hich range o v er heads, A ’ -positions, and A -positions. A simp lified f ormulation of R elativized Minimality is giv en in ( 36 ). (36) R elativized Minimality (Rizzi 1990 ; a v ersion tak en fr om Rizzi 2011 , p.221-222) a. In the configuration ... X ... Z ... Y ..., a local r elation cannot connect X and Y if Z interv enes and Z is of the same structural type as X. b. S tructural types: (i) A′ positions, (ii) A positions, and (iii) heads. 3.2.2 The parti cular natur e of the HM C and e x cepti ons to the HM C W hile R elativized Minimality appears to successfull y incorporate the HM C into a mor e general min- imality condition in language, it r emains unclear w h y the thr ee structural types in ( 36 b ) should f orm a natural class. Rizzi ( 2001 , p.90-91) points out that the typology of structural types in v ol v es “ at least tw o irr educible distinctions. ” (37) T w o distinctions in structural types a. a distinction betw een heads and p hrases, and in the latter class, b. a distinction betw een positions of arguments (A -positions ) and of non-arguments (A′-positions ). In other w or ds, R elativized Minimality still imp licates that head mo v ement is substantiall y dif - f er ent fr om p hrasal mo v ement, in the sense that w hile a head is generall y an interv ener of head mo v e- ment, a p hrase is not al w a y s an interv ener of p hrasal mo v ement ( as it depends on w hether a position is argument - r elated). S ubsuming the HM C under R elativized Minimality does not r emo v e the par - 37 3.2. INTER VENTI O N EFFECTS AND HEAD M O VEMENT ticular natur e of the HM C: head mo v ement is specificall y constrained b y stricter locality conditions compar ed to p hrasal mo v ement. S ubsequent discussions and r ef ormulation of R elativized Minimal - ity f ocus on the f eatural encoding of potential p hrasal interv eners (Chomsky 1995b ; Rizzi 2001 , 2004 ). H ead mo v ement and the HM C r eceiv e r elativ el y little discussion in the study of interv ention. 8 I str ess that the particular natur e of the HM C does not lie in w hether they can be deriv ed fr om mor e general princip le of the grammar 9 , but in the suggestion that they r epr esent a gener al constraint on mo v ement of a particular structural type. I t is true that some instances of p hrasal mo v ement ma y disp la y similar eff ects, but the lack of generality marks a significant diff er ence betw een the locality constraints on head and p hrasal mo v ement. Empiricall y , the HM C appears to be too strict as a gener al locality constraint on head mo v ement. E vidence r ev eals that ther e ar e man y cases that constitute a violation to the HM C. The evidence comes r oughl y in tw o main types. 10 The first one concerns a construction w her e a v erb/ particip le is argued to be fr onted bef or e an interv ening auxiliary v erb. This construction is often r ef err ed to as Long H ead M ovement (LHM, Lema and Riv er o 1990 ; Riv er o 1991 ; 1993 ; 1994 ; R oberts 1994 ; W ilder 1994 ; Borsley , Riv er o, and S tep hens 1996 , among others ). 11 T w o examp les fr om Bulgarian and B r eton ar e giv en in ( 38 ). 8. But see K oopman ( 1984 ) and Y . Li ( 1990 ) f or discussions on an A - v s. A ’ -distinction on heads. See also R oberts ( 2001 ) f or an operator v s. non-operator distinction on heads. 9. F or examp le, Chomsky ( 1986 ), Bak er ( 1988 ), and Rizzi ( 1990 ) suggest that the HM C can be deriv ed fr om the Empty Category Princip le (ECP ). Bak er ( 1988 ) also suggests that the ban of ex corporation ma y be due to illicit trace within w or d. 10. I f ocus on cases in the v erbal domain, but violation to the HM C has also been argued to be attested in the nominal domain, such as clitic climbing in R omance languages; see R oberts ( 1991 , 2010 ) f or discussions. 11. These cases earn their name as “long” in the sense that they ar e not strictl y local, as opposed to “Short H ead M o v e- ment” in Germanic languages that do not in v ol v e a v erb cr ossing an auxiliary . N ote that LHM is r eported to be clause- bounded, and not to be confused with long -distance dependencies that ma y occur acr oss clauses. 38 3.2. INTER VENTI O N EFFECTS AND HEAD M O VEMENT (38) Examp les of Long H ead M o v ement a. Bulgarian (Lema and R iv er o 1990 , p.334) Pr očel r ead sum ha v e+pres+1sg knigata book+the ‘I ha v e r ead the book ( comp letel y ). ’ b. Br eton (Borsley , Riv er o, and S tep hens 1996 , p.53) Lennet r ead en 3sg.masc deus has Y ann Y ann al the levr book ‘ Y ann has r ead the book. ’ I t is r eported that similar constructions ar e f ound in O ld R omance, E ur opean P ortuguese, M od- ern R omanian and man y Sla vic languages (Bulgarian, Serbo-Cr oatian, Czech, etc.) (Lema and Riv er o 1990 ). Generall y , LHM is said to disp la y r oot eff ects, i.e., it is not a v ailable in embedded contexts. Languages ma y diff er in the inflectional f orm of the fr onted v erb – they ma y be in infinitiv e f orms or in particip le f orms. Languages ma y also diff er in terms of the trigger and the interpr etiv e eff ects. W ithout going into the details of language-specific pr operties of LHM, these cases typicall y in v ol v e fr onting of the v erb o v er the auxiliary v erb. This constitutes a dir ect violation to the H ead M o v ement Constraint. The general schema can be r epr esented in ( 39 ). (39) A schematic r epr esentation of LHM CP V A uxP A ux VP t V ( object ) The other type of empirical evidence against the strictl y local natur e of head mo v ement comes fr om pr edicate cleft or v erb fr onting constructions. T ypicall y , in these constructions, an additional cop y of the v erb appears on its o wn in the initial/ perip hery position. These constructions ar e r eported 39 3.2. INTER VENTI O N EFFECTS AND HEAD M O VEMENT to con v ey discourse eff ects such as a f ocus r eading or a topic r eading of the v erb, subject to language v ariations. The tw o copies of the same v erbs can also be separated b y clausal boundaries, establishing long distance dependencies. I giv e tw o examp les fr om V ata and Spanish in ( 40 ), w her e the tw o v erbs ar e separated b y a clausal boundary . (40 ) Examp les of pr edicate cleft/ v erbal fr onting constructions a. V ata (K oopman 1984 , p.159) yī come IJ O s/he w à w ant [ nā N A à w e yī come ] ‘S/he w ants us to come. ’ b. Spanish (V icente 2007 , p.79) Compr ar , buy .inf J uan J ha has dicho said [ que that M aría M ha has compr ado bought un a libr o book ] ‘ As f or buying, J uan has told me that M aría has bought a book. ’ I t has been argued that the tw o v erbs ar e r elated b y head mo v ement and the mo v ement is sensitiv e to syntactic islands (K oopman 1984 ; V icente 2007 ). If this is the case, then w hen the v erb mo v es to the initial position , it cr osses at least a head along its path. These cases constitute violations to the HM C. 12 S imilar beha viors ar e r eported in pr edicate cleft/ v erb fr onting constructions in man y other languages. A non-exhaustiv e list is giv en in ( 41 ). All these cases ar e argued to in v ol v e v erb/head mo v ement. (41) A non-exhaustiv e list of pr edicate cleft/ v erb fr onting constructions 13 a. V ata (K oopman 1984 ) b. Bulgarian (Lambo v a 2004 ; H arizano v 2019 ) c. H ebr ew (Landau 2006 ) 12. The doubling eff ects of the v erb ar e argued to be due to independent factors. 13. F or extensiv e discussions on cr oss- linguistic v erb/ v erbal fr onting constructions, see H ein ( 2018 ). 40 3.2. INTER VENTI O N EFFECTS AND HEAD M O VEMENT d. H ungarian (Ü r ög di 2006 ; V icente 2007 ) e. Spanish (V icente 2007 ) f. H aitian (H arbour 2008 ) g. N upe (Kandybo wicz 2008 ) h. V ietnamese (T rinh 2009 ) i. B razilian P ortuguese (Ba stos-Gee 2009 ) j. M andarin (Cheng and V icente 2013 ) k. Asante T wi (H ein 2018 ) l. R ussian (Antonenk o 2019 ) m. F innish (B rattico 2021 ) As such, the cases of LHM and pr edicate cleft/ v erb fr onting indicate that the HM C ar e too strict as a general locality constraint on head mo v ement. On a w eak thesis, the HM C must be r evised such that it operates onl y on selected domains of head mo v ement (in a w a y similar to a f eatural -based R elativized Minimality ) On a str ong thesis, the HM C ma y be non-existent in the first p lace, if the local natur e of head mo v ement (in the cor e cases in support of the HM C) ma y be deriv ed fr om other princip les in the grammar . I t should be r emark ed that cases of head mo v ement as in LHM and pr edicate cleft/ v erb fr onting ar e not entir el y unconstrained. F or examp le, LHM ( in Bulgarian and R umanian ) cannot mo v e acr oss negation, as in ( 42 ). 14 Pr edicate clefts in V ata cannot span acr oss a comp lex nominal, as in ( 43 ). These examp les sho w that head mo v ement is not ex ceptionall y unconstrained in terms of locality . 14. See R oberts ( 2001 ) f or a potential exp lanation based on the distinctions of operator heads and non-operator heads. 41 3.2. INTER VENTI O N EFFECTS AND HEAD M O VEMENT (42) LHM in Bulgarian cannot cr oss negtion * (Lema and Riv er o 1990 , p.337) Pr očel r ead ne not sum ha v e+pres+1sg knigata book+the Int.: ‘I ha v e r ead the book ( comp letel y ). ’ (43) P r edicate cleft in V ata cannot span acr oss a comp lex nominal * (K oopman 1984 , p.159) tākā sho w ǹ y ou w à lik e [ f òt IJ o pictur e m Ū m IJ U ITIT ǹ y ou tākā -á IJ O sho w ed-rel àbà ] Aba Int.: ‘ Y ou lik e the pictur e y ou sho w ed to Aba. ’ In sum, empirical evidence speaks against the HM C as a general constraint on head mo v ement and calls f or a mor e fine-grained locality theory of head mo v ement that is not ex clusiv el y f ormulated with r ef er ence to structural types. 3.2.3 Base gener ati on and r emnant mo v ement and as alternati v es Eff orts ha v e been made in the literatur e to determine the corr ect anal y sis of LHM and pr edicate cleft/ v erb fr onting. If it can be sho wn that in these cases no v erb/head mo v ement is in v ol v ed, then the HM C can be maintained. In w hat f ollo w s, I briefl y discuss tw o potential alternativ es to a head mo v ement anal y sis. The first one is a base generation appr oach. F or LHM cases, Embick and Izv orski ( 1997 ) suggest that the particip le-auxiliary w or d or der is not due to syntactic head mo v ement but the enclitic natur e of ( certain ) auxiliaries. In other w or ds, the v erb is base-generated in the lo w position, and the alter - nation in w or d or der is due to a morp hological merger that operates on heads. Th e idea is illustrated in ( 44 ). 42 3.2. INTER VENTI O N EFFECTS AND HEAD M O VEMENT (44) An illustration of the base generation anal y sis of Bulgarian P articip le- A uxiliary or der a. The v erb base-generat es at a lo w position: [ A ux [ V Obj ] ] b. A morp hological merger app lies to affix the auxiliary to the particip le: V - A ux Obj ( w her e A ux is an enclitic ) A base generation anal y sis r epr esents a potential alternativ e to a head mo v ement appr oach. 15 If this is the case, no head mo v ement is needed to deriv e the r elev ant or der . On a similar v ein, a base generation anal y sis has been pr oposed f or pr edicate cleft/ v erb fr onting in Y iddish and B razilian P ortuguese in Cable ( 2004 ). Cable ( 2004 ) suggests that pr edicate cleft in Y iddish in v ol v es a base generated topic, w hich can be lexicall y non- identical to its v erbal associate. In ( 45 ), the v erb f orn ‘tra v el’ is argued to be base-generated in the initial position, as it does not corr espond to a g ap or a cop y in the lo w er clause. 16 (45) A base generation anal y sis of Y iddish pr edicate cleft (V icente 2007 , p.82; cf. Cable 2004 , p.9) F orn tra v el.inf bin am ikh I geflo y gn flo wn k eyn to N yu - Y ork N ew - Y ork ‘ As f or tra v eling, I’v e fl o wn to N ew Y ork. ’ Another alternativ e to a head mo v ement anal y sis is to posit r emnant VP mo v ement. In German, P articip le- A uxiliary or der is possible as with man y cases of LHM. F or examp le, den Besten and W e- belhuth ( 1990 ) argue that instead of the v erb mo ving on its o wn, the surface or der r esults fr om VP mo v ement. This VP , ho w ev er , is a r emnant cr eated b y object scrambling. The object has been mo v ed out of the VP bef or e VP mo v ement. The VP thus contains onl y the v erb. 17 W hen it mo v es, it appears 15. See F iantis ( 1999 ) f or another p honological alternativ e to LHM in Bulgarian; see also H arizano v ( 2019 ) f or a def ense of a head mo v ement anal y sis. 16. Indeed, Cable ( 2004 ) suggests that the initial v erbal topic is base-generated in the minimal clause and ma y subse- quentl y mo v es to its surface position in the perip hery . If so, head mo v ement is still needed. 17. F anselo w ( 2002 ) argues that, w hile r emnant mo v ement ma y be independentl y needed in other contexts in German, it has a number of shortcomings compar ed to a v erb raising alternativ e in cases w her e a bar e v erb appears in the initial position. 43 3.2. INTER VENTI O N EFFECTS AND HEAD M O VEMENT that the v erb is mo ving on its o wn. T o illustrate, the P articip le- A uxiliary or der in ( 46 ) in v ol v es object scrambling ( out of VP ) and subsequent VP mo v ement to the initial position. (46) A r emnant mo v ement anal y sis of P articip le- A uxiliary or der in German [ VP t i Gelesen ] r ead hat has H ans H ans [ das the Buch ] i book nicht not t VP ‘H ans has not r ead the book. ’ S u ch a r emnant mo v ement appr oach has subsequentl y be app lied to pr edicate cleft/ v erb fr onting cases and other r elev ant structur es (K oopman 1997 ; K oopman and S zabolcsi 2000 ; M ahajan 2003 ; N i lsen 2003 ; M üller 2004 , among man y others ). F or examp le, K oopman ( 1997 ) abandons her head mo v ement anal y sis of V ata pr edicate cleft and pr oposes instead a r emnant mo v ement appr oach in a w a y similar to the German case. The discussion in the subsection indicates that if the the HM C- violating cases of head mo v ement can be alternativ el y anal yzed as base generation or r emnant mo v ement, then the HM C can be main- tained as a general constraint on head mo v ement. 3.2.4 Interim summary In summary , setting aside the particular natur e of the HM C w hen compar ed to p hrasal mo v ement, the empirical v alidity of the HM C hinges on the corr ect anal y sis on the cases such as LHM and pr ed- icate cleft/ v erb fr onting. In w hat f ollo w s, I pr esent evidence fr om Cantonese and argue that a head mo v ement appr oach is necessary to captur e the empirical pr operties of the f our v erb doubling con- structions. Cruciall y , I sho w that these constructions ar e not subject to the HM C, and thus the HM C cannot be maintained as a general constraint on head mo v ement. H o w ev er , I also sho w that head mo v ement is not particularl y unconstrained. I sho w that it exhibits similar (f ocus ) interv ention ef - f ects as observ ed with p hrasal mo v ement. 44 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS 3.3 V erb doub ling constructi ons and discour se eff ects In this section, I turn to v erb doubling cases in Cantonese. In view of the little attention on these constructions in the literatur e, I pr esent a detailed description of both the syntactic and semantic pr operties of these v erb doubling constructions in Cantonese. The discussion serv es as the empirical gr ounding f or the pr oposed head mo v ement anal y sis. 3.3.1 T ypes of v erbs V erb doubling constructions ar e pr oductiv e in Cantonese. I illustrate the v ariety of v erbs compatible with v erb doubling constructions. W hile the examp les giv en in ( 28 ) all in v ol v e a transitiv e v erb, I sho w that transitivity of the doubled v erbs does not aff ect the acceptability of the v erb doubling construc - tions, on a par with semantic pr operties such as the stativ e/ ev entiv e distinction and p honological pr operties such as the number of sy llables. 18 In topic constructions of v erbs, the doubled v erb can be an intransitiv e ( unergativ e ) v erb, such as siu ‘laugh ’ , or a disy llabic ( monomorp hemic ) v erb, such as pisen ‘pr esent’ ( w hich is an English loan- w or d), as sho wn in the sentences in ( 47 ). (47) T opic constructions of v erbs a. S iu laugh Aaming Aaming hai co p siu -dak laugh-res hou v ery daaiseng. loud Batgw o but k eoi he gongje speak hou v ery saiseng. lo w . v oice ‘ As f or laughing, Aaming laughs loudl y . But he speaks v ery softl y . ’ 18. W henev er appr opriate, I include possible continuations after the target sentence to facilitate naturalness. 45 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS b. Pisen pr esent Aaming Aaming hai co p pisen -gw o pr esent -exp ni -pin this-cl man. paper Batgw o but k eoi he jinghai onl y gong - zo talk -perf ng fiv e fanzung. minute ‘ As f or pr esenting, Aaming has pr esented this paper . But he onl y talk ed f or fiv e minutes. ’ In ‘ ev en ’ - f ocus constructions, sentences in ( 48 ) sho w that the targeted v erb can be an unaccusativ e v erb such as zoek ‘be.turned.on ’ or a disy llabic transitiv e v erb such as daamsam ‘w orry’ . (48) ‘E v en ’ - f ocus constructions of v erbs a. Lin ev en zoek on bou cl dinnou computer dou also m- zoek . not -on Zanhai r eall y w aai -dak br ok en-res hou v ery citdai. entir e ‘This computer cannot ev en B E TURNED O N . (I t has ) r eall y br ok en do wn comp letel y . ’ b. Lin ev en daamsam w orry Aaming Aaming dou also m- daamsam not - w orry nei. y ou M - hou not -good w aa sa y ziugu tak e.car e nei. y ou ‘ Aaming does not ev en W O RR Y about y ou, not t o mention tak e car e of y ou. ’ P aris ( 1998 ) observ es that in M andarin stativ e pr edicates lik e ‘tir ed’ and ‘ cheap ’ cannot be targeted in ‘ ev en ’ - f ocus constructions, as sho wn in the Cantonese counterpart in ( 49 a ). M atthew s and Y ip ( 1998 ) also observ es a similar constraint in Cantonese, as in ( 49 b ). (49) Illicit stativ e pr edicates in ‘ ev en ’ - f ocus constructions a. ( cf. P aris 1998 , p.146) ?? Lin ev en gui tir ed k eoi s/he dou also m- gui . not -tir ed ‘S/he is not ev en tir ed. ’ 46 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS b. * G w ai expensiv e dou also m- gw ai . not -expensiv e J anjan ev ery one dou all maai -dak - hei. buy -able-up (M atthew s and Y ip 1998 , p.184) ‘(I t is ) not ev en expensiv e. E v ery one can aff or d (it ). ’ H o w ev er , ther e ar e stativ e pr edicates that can be f ocused in ‘ ev en ’ - f ocus constructions, such as zi ‘kno w’ and jit ‘hot’ , as sho wn in ( 50 ). I ther ef or e suggest that ther e is no general ban on stativ e pr edicates in ‘ ev en ’ - f ocus constructions, but f ocusing stativ e pr edicates ma y r equir e a specific context to sound natural. (50 ) Licit stativ es pr edicates in ‘ ev en ’ - f ocus constructions a. N i -gin this-cl si ev ent lin ev en zi KN O W k eoi s/he dou also m- zi . not - kno w M - hou not -good w aa sa y bei giv e jigin. comment ‘S/he did not ev en kno w this, not to mention comment ( on it ) . ’ b. Lin ev en jit hot go cl faanhap lunchbo x dou also m- jit . not - hot N ei y ou giu tell jan person dim ho w sik? eat ‘The lunchbo x (i.e., the f ood in the lunchbo x) is not ev en H O T . H o w can I eat (it )?’ Copula f ocus constructions of v erbs disp la y similar flexibility in the choice of v erbs. S ince the constructions, as far as I kno w , ha v e not been documented in the literatur e, I illustrate the construc - tion with f our examp les and pr o vide a context f or each of them. (51) Copula f ocus constructions of v erbs a. Context: A J apanese teacher is ta lking about one o f his student, Aaming: H ai co p gong speak Aaming Aaming gong - m-dou sa y - not -able tongseon fluent ge m o d jatman J apanese ze1. sfp K eoi he se-dak write-res hou v ery hou. good ‘ Aaming cannot SP EAK fluent J apanese onl y . H e writes v ery w ell. ’ 47 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS b. Context: A fr equent tr aveler is ta lking about his attitude towar ds J apan: H ai co p zyu liv e ngo I m-soeng not - w ant zyu liv e hai in J atbun J apan ze1. sfp N g o I bunsan self hai co p zungji lik e heoi go J atbun J apan leoihan tra v el ge3. sfp ‘I don ’t w ant to LIVE in J apan onl y . I lik e to tra v el to J apan. ’ c. Context: A r ecruitment manager is talking about one o f the interviewees, Aaming: H ai co p l eng pr etty Aaming Aaming m-gau not -enough Aafan Aafan l eng pr etty ze1. sfp K eita other je thing dou all hai be Aaming Aaming sikhap suitable di. a.bit ‘ Aaming is not as P RETTY as Aafan onl y . F or other things, Aaming is mor e suitable. ’ d. Context: A pr o f essor is talking a bout one o f his students, Aaming: H ai co p pisen pr esent k eoi s/he mou not.ha v e seonsam confidence pisen -dak pr esent -res hou good ze1. sfp Se write peipaa paper gew aa if k eoi s/he gok dak think se-dou. write-able ‘S/he lacks the confidence to P RESENT w ell onl y . If If ( s/he is ) to write paper , s/he thinks ( s/he ) can write (it ). ’ Lastl y , f or dislocation cop ying of v erbs, possible v erbs range fr om transitiv e v erbs to the copula v erb and ( disy llabic ) attitude v erbs, illustrated in ( 52 ). 19 19. M atthew s and Y ip ( 1998 ) notes that sentences lik e ( 52 a ), w her e an wh -expr ession is in v ol v ed, con v ey s an “idiomatic” meaning, though admitting that it is pr oductiv e in colloquial Cantonese. The “idiomatic” meaning can be r egar ded as a rhetorical meaning or a touch of w hining f or ce that is attributed to the wh -expr essions (T sai 2021 ). 48 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS (52) Dislocation cop ying of v erbs a. (M atthew s and Y ip 1998 , p.186) Coeng sing matje w hat aa3 sfp coeng ? sing ‘W hat ar e ( y ou ) singing?’/ ‘W h y ar e ( y ou ) sing?’ b. (L. Y .-L. Cheung 1997 , p.9) H ungsaudou karate wui club hai co p mou not.ha v e je thing hok learn go3 sfp lo1 sfp hai . co p ‘The karate club off ers nothing f or us to learn. ’ c. (K. K. Chan 2016 , p.19) N go I go k dak think nei y ou zoengloi futur e wui will hou v ery meimun fruitful aa3 sfp go k dak . think ‘I think y our futur e will be v ery fru itful. ’ So far , all the abo v e examp les sho w that v erb doubling constructions ar e pr oductiv e. Ther e is, ho w ev er , a r estriction on the choice of v erbs. F or examp le, some modal v erbs cannot be targeted in the topic constructions and the tw o f ocus constructions. H o w ev er , dislocation cop ying is able to target these v erbs. Compar e ( 53 ) and ( 54 ): (53) Some modal v erbs cannot be targeted in topic and f ocus constructions a. ?? H onang possible Aaming Aaming hai co p honang possible m- lai not -come ge2. sfp ‘ As f or w hether it is possible, it is poss ible that Aaming does not come. ’ b. ?? Lin ev en wui will Aaming Aaming dou also m- wui not - will heoi. go Int.: ‘ Aaming is not going - ther e is ev en no possibility that he will go. c. ?? H ai co p jinggoi should Aaming Aaming m- jinggoi not -should heoi go ze1. sfp Int.: ‘ Aaming SH O ULD not go onl y . ’ 49 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS (54) Dislocation cop ying of modal v erbs a. Aaming Aaming honang possible m- lai not -come laa3 sfp honang . possible ‘I t is possible that Aaming is not coming. ’ b. K eoi s/he wui will zoeng d is gaan cl uk house maai -ceot - heoi sell -out -go gaa3 sfp laa3 sfp wui . will (L. Y .-L. Cheung 2015 , p.248) ‘S/he will sell the house. ’ c. (Lai 2019 , p.259) N ei y ou jinggoi should tung with jan person gong talk ge2 sfp jinggoi . should ‘ Y ou should tell others. ’ As f or w h y dislocation cop ying of v erbs is allo w ed to target a larger set of v erbs, I suggest that the diff er ence lies in the discourse eff ects br ought along b y these constructions. Anticipating the dis- cussion of their corr esponding discourse eff ects in § 3.3.3 , the topic and f ocus constructions in v ol v e contrastiv e interpr etation of the v erbs in the perip hery , w her eas dislocations cop ying diff ers in mark - ing the v erb as def ocused or giv en. I suggest to connect the sp lit in discourse eff ects to the range of possible targets in these constructions. I suggest that onl y v erbs that denote a pr oper ev entuality can be contrasted or f ocused, w her eas all v erbs can be def ocused. 20 This suggestion is in line with the observ ation that dislocation cop ying can ev en target adv erbial elements that cannot be topicalized or f ocused. These examp les include jiging ‘ alr eady’ and dou ‘ all. ’ (55) Dislocation cop ying of adv erbial elements a. (L. Y .-L. Cheung 2015 , p.234) K eoi s/he jiging alr eady heoi -gw o go-exp laa3 sfp jiging . alr eady ‘S/he has alr eady been ther e. ’ 20. This is supported b y the observ ations that auxiliary v erbs fail to be targeted in pr edicate clefting in Spanish (V icente 2007 ) and particip le fr onting in Bulgarian (H arizano v 2019 ). 50 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS b. (L. Y .-L. Cheung 2015 , p.236) K eoi s/he dou also lai come gaa3 sfp dou . all ‘S/he also comes. ’ As a final r emark, the v erb in these v erb doubling constructions cannot be r egar ded as a nomi - nalized v erb, a possibility mentioned in passing in Sh yu ( 1995 , p.14, fn.11). 21 This is because v erbs cannot appear after jau ‘ha v e’ in existential constructions ( see, e.g., H uang 1987 ) or after the f ocus- marking dak ‘ onl y’ (T ang 2002 ), both of w hich can onl y tak e nominal elements . If v erbs w er e able to be nominalized, w e w ould expect that they could appear in constructions that can onl y target nomi - nals, contrary to facts. I ther ef or e maintain that the doubled elements in v erb doubling constructions ar e genuine v erbs but not deriv ed nominals. (56) Existential constructions with jau ‘ha v e’ cannot target v erbs a. J au ha v e jat -bun one-cl s yu book ngo I soeng w ant maai. buy . ‘Ther e is a book that I w ant to buy . ’ b. * J au ha v e maai buy ngo I soeng w ant maai buy jat -bun one-cl syu. book Int.: ‘Ther e is a book that I w ant to buy . ’ (57) F ocus constructions with dak ‘ onl y’ cannot target v erbs a. Dak onl y H oenggong H ong.K ong ngo I zyu -gw o. liv e-exp ‘I ha v e onl y liv ed in H ong K ong. ’ 21. This possibility is not imp lausible. F or examp le, pr edicate cleft in Bùlì is obligatoril y mark ed b y an o v ert nominal - izing suffix (Hiraiw a 2002 ). 51 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS b. * Dak onl y zyu liv e ngo I zyu -gw o liv e-exp H oenggong. H ong.K ong Int.: ‘I ha v e onl y liv ed in H ong K ong. ’ 3.3.2 M orp ho-s yntacti c pr operti es and v ariants 3.3.2.1 T opi c constructi ons of v erbs In topic constructions of v erbs, ther e is no obligatory morp ho-syntactic marking on the topicalized v erb. A v erb appears in its bar e f orm. W hile the lo w er v erb is often associated with the copula, it is not obligatory either , especiall y in shorter sentences. N ote that I assume with C. C.-H. Cheung ( 2015 ) and Erlewine ( 2020b ) that the copula v erb hai serv es as a f ocus particle instead of a genuine v erb, w hen it is used to mark f ocus position (f or discussions, see § 3.3.2.3 ). (58) The copula bef or e the lo w er v erb is optional Zou do k eoi s/he (hai) co p wui will zou do ge2. sfp ‘ As f or doing, s/he will do. ’ T ypicall y , the construction has a concessiv e character (M atthew s and Y ip 1998 ). I t usuall y comes with the sentence- final particle ge2 , w hich con v ey s concession or r eserv ation. Alternativ el y , a but - clause in the continuation is pr ef err ed. N ote that the concessiv e sense does not disappear ev en in the absence of ge2 or a but -clause, as pointed out b y M atthew s and Y ip ( 1998 , p.179). (59) The sentence- final particle ge2 is optional J am drink k eoi s/he wui will jam . drink Batgw o but m- wui not - will jam drink hou v ery dou. much ‘ As f or drinking, s/he will drink. But s/he will not drink much. ’ The schematic pattern of topic constructions of v erbs is summarized in T able 3.2 . 52 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS T ype Left P erip hery S M ark er V base O SFP Right P erip hery T opic - V V S ( hai ) V O ( ge2 ) - T able 3.2: The schematic pattern of topic con structions of v erbs N ote that in the discussions of v erb fr onting in M atthew s and Y ip ( 1998 ) and M atthew s and Y ip ( 2011 , p.88-89), they pr esent cases with zau ‘then ’ ( ex cept a f ew examp les with dou ‘ all’ , w hich is an instance of ‘ ev en ’ - f ocus constructions, under the classification in this chapter ). F or examp le, (60 ) T opic construciotns of v erbs with zau ‘then ’ (M atthew s and Y ip 2011 , p.89) Leng pr etty zau then l eng. pr etty Batgw o but zau then gw ai - zo expensiv e-perf di. a.bit ‘(If it is about w hether it is ) pr etty , then it is pr etty . But (it is ) a bit expensiv e. ’ I ex clude these cases in the discussions f or a f ew r easons. F irst, these sentences ma y in v ol v e a bi - clausal structur e, since zau ‘then ’ signals the pr esence of a conditional clause. 22 The doubled v erb ma y be a significantl y r educed f orm of a conditional clause, con v eying a non-contrastiv e topic r eading, e.g., ‘if it is about w hether it is beautiful ’ in ( 60 ). 23 Second, the initial v erb need not ha v e an exact cop y in the original clause, as in ( 61 ). (61) N o doubling eff ects with zau ‘then ’ Context: in r eply to t he question o f whether the speaker knows a lot o f Chinese. (M atthew s and Y ip 2011 , p.88- 89, adapted) Gong speak zau then hai. be T ai r ead zau than m- hai. not -be ‘(If it is about ) speaking, the n y es. (If it is about ) r eading, then no. ’ As will be discussed in gr eater details in § 3.4.1 , lexical identity eff ects ar e crucial to diagnose syn- tactic dependencies. F or the r est of the chapter , I f ocus on cases with doubling eff ects onl y . 22. This possibility is also mentioned in M atthew s and Y ip ( 1998 ), but they suggest that this is the case f or v erb doubling in imperativ e sentences. I suggest that this app lies to all zau -sentences. 23. This is in line with the suggestion in Liu ( 2004 ) f or M andarin and Shanghainese that v erb doubling ( or identical topics in his term ) ma y be grammaticalized fr om r educed conditionals. 53 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS 3.3.2.2 ‘E v en ’- f ocus constructi ons of v erbs T urning to ‘ ev en ’ - f ocus constructions, despite its name, the f ocus particle lin ‘ ev en ’ can be dr opped. 24 W hat is obligatory in these constructions is the adv erb dou , in addition to the v erb doubling eff ects. (62) The obligatory dou and the optional lin (Lin ) ev en tai r ead aaming Aaming *( dou ) also m- tai not - r ead ni -bun this-cl syu book w o4. sfp ‘ Aaming didn ’t ev en READ this b ook ( to m y surprise ). ’ The construction is compatible with diff er ent sentence- final particles, such as the mirativ e wo4 that con v ey s speak ers’ surprise, and the questions particle me1 . I t is also compatible with their absence ( a s in ( 28 b )). (63) Compatibility with diff er ent SFP s Lin ev en tai r ead Aaming Aaming dou also m- tai not - r ead ni -bun this-cl syu book me1? sfp .q ‘I s it that case that Aaming didn ’t ev en READ this book?’ The schematic pattern of ‘ ev en ’ - f ocus constructions of v erbs can be summarized as f ollo w s. The notation ( an y sfp) in the table indicates its optionality and unselectivity . T ype Left P erip hery S M ark er V base O SFP Right P erip hery ‘E v en ’ - V ( lin )- V S dou V O ( an y sfp) - T able 3.3: The schematic pattern of ‘ ev en ’ - f ocus constructions of v erbs N ote that as with the object/ p hrasal counterparts, the lin - mark ed v erb can appear in a clause- internal position after the subject, in addition to the clause- initial position. Thr oughout the discus- sion, I will not addr ess the potential diff er ences betw een these tw o positions of ‘ ev en ’ - f ocus (f or dis- 24. The categorial status of lin ‘ ev en ’ is not uncontr o v ersial. H er e I assume Cantonese lin patterns with M andarin lian in being a f ocus particle instead of a v erb or other lexical categories (T sai 1994 ; Sh yu 2004 ; Badan 2007 ). 54 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS cussions, see Sh yu ( 1995 ), among others ). (64) Clause- internal lin is possible Aaming Aaming lin ev en dim touch dou also m-gam not -dar e dim tounch ni - zek this-cl dungmat. animal ‘ Aaming doesn ’t ev en dar e t o T O U CH this animal. ’ 3.3.2.3 Copula f ocus constructi ons of v erbs The signatur e pr operty of copula f ocus constructions of v erbs is the obligatory pr esence of hai in the clause- initial position. 25 (65) Obligatory hai *(H ai) co p gong speak Aaming Aaming m-sik not - kno w gong speak jatman J apnese ze1. sfp ‘ Aaming does not kno w ho w to SP EAK J apanese onl y . ’ Copula f ocus constructions should not be conflated with the mor e discussed hai...ge3 cleft con- structions ( the M andarin counterpart of shi...de cleft constructions ). As observ ed in L. L. S. Cheng ( 2008 ), the copula f ocus constructions (= her bar e- shi sentences ) ar e incompatible with the sentence- final de in M andarin, or ge3 in Cantonese. (66) Incompatible with sentence- final ge3 * H ai co p gong speak Aaming Aaming m-sik not - kno w gong speak jatman J a pnese ge3. sfp Int.: ‘ Aaming does not kno w ho w to SP EAK J apanese onl y . ’ Ther e is a r equir ement f or the choice of sentence- final particles. The sentence is degraded in the 25. N ote that the categorial status of the copula v erb is a matter of debate. C. C.-H. Cheung ( 2015 ) and Erlewine ( 2020b ) argues that it is a f ocus mark er , w her eas L. L. S. Cheng ( 2008 ) and P an ( 2019 ) suggest that it is a genuine copula v erb ( and in v ol v es a bi -clausal structur e ). The distinction does not bear on the discussion. 55 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS absence of r estrictiv e/ f ocus- r elated particles such as ze1 or zaa3 ( cf. F ung 2000 ). 26 (67) Obligatory r estrictiv e sentence- final particles H ai co p gong speak Aaming Aaming m- zik not - kno w gong speak jatman J a apnese ??( ze1/ sfp zaa3). sfp ‘ Aaming does not kno w ho w to SP EAK J apanese onl y . ’ The schematic pattern of copula f ocus constructions of v erbs is summarized belo w . T ype Left P erip hery S M ark er V base O SFP Right P erip hery Copula- V hai - V S - V O ze1 / zaa3 - T able 3.4: The schematic pattern of copula f ocus constructions of v erbs S imilar to the case of topic constructions of v erbs discussed in § 3.3.2.1 , ther e appears to be a v ari - ant of copula f ocus constructions w hich in v ol v es zau ‘then ’ . I ( again ) suggest that these cases ma y in v ol v e a r educed v erbal conditional clause in a bi -clausal structur e. Als o, ther e ar e no lexical identity eff ects, as illustrated in ( 68 ). I ex clude this type of constructions in the discussion and f o cus on the v erb doubling cases. (68) Copula f ocus constructions with zau ‘then ’ a. H ai co p zau go nei y ou zau then faaidi quickl y zau go laa1. sfp ‘(If it is decided to ) go, then y ou should go qu ickl y . ’ 26. This is diff er ent fr om the clause- internal hai , w her e ther e is no such r equir ement on sentence- final particles. N ote also its compatibitlity with ge3 . (i) Aaming Aaming hai co p m-sik not - kno w gong speak jatman J apanese ( ze1/ sfp / zaa3 sfp / / sfp ge3). ‘ Aaming ( onl y ) doesn ’t kno w ho w to speak J apanese. ’ 56 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS b. H ai co p zau go nei y ou zau then mai not lam think gam too do much laa1. sfp (If it is decided to ) go, then y ou should not think t oo much. ’ 3.3.2.4 Disl ocati on cop ying of v erbs The defining pr operty of dislocation cop ying of v erbs is the final position of the doubled v erb. In addition, sentence- final particles ar e obligatory in the construction, this app l ying to cases of right dislocation/ dislocation cop ying in general (L. Y .-L. Cheung 1997 ; Lai 2019 ; K.-F . Y ip 2020 ). (69) Sentence- final particles ar e obligatory in dislocation cop ying of v erbs N go I jiu need lok do wn gaai str eet maai buy di cl je thing *( aa3) sfp jiu . need ‘I need to go out to buy some thing. ’ One r elated v ariant of dislocation cop ying of v erbs is that the base v erb ma y be dr opped, and thus the doubling eff ects ar e not obligatory . This contrasts with the pr evious thr ee constructions. (70 ) Right dislocation/ dislocation cop ying of v erbs a. N go I ( hai ) be H oenggong H ong.K ong jan person aa3 sfp hai . be ‘I am a H ongk onger . ’ b. Aaming Aaming tingjat tomorr o w ( hoji ) ma y lai come aa3 sfp hoji ma y ‘ Aaming ma y come tomorr o w . ’ Right dislocation of v erbs of this kind is specificall y discussed in Lee ( 2017 , 2021 ). I suggest that the tw o constructions ar e the same constructions and optionality in doubling arises as a consequence of p honological linearization ( see Chapter 5 f or further discussions ). This issue, ho w ev er , should not 57 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS concern us in the curr ent discussion of the locality issues of head mo v ement. In the r est of this chapter , I f ocus on dislocation cop ying of v erbs. The schematic pattern of dislocations cop ying of v erbs is r epr esented belo w . T ype Left P erip hery S M ark er V base O SFP Right P erip hery DC- V - S - ( V ) O an y sfp V T able 3.5: The schematic pattern of dislocation cop ying of v erbs 3.3.2.5 Interim summary The pr operties of the v erb doubling constructions discussed this far ar e summarized in T able 3.6 . T ype Left P erip hery S M ark er V base O SFP Right P erip hery T opic - V V S ( hai ) V O ( ge2 ) - ‘E v en ’ - V ( lin -) V S dou V O ( an y sfp) - Copula- V hai - V S - V O ze1 / zaa3 - DC- V - S - ( V ) O an y sfp V T able 3.6: The sch ematic patterns of v erb doubling constructions In terms of morp ho-syntactic marking on the v erb in the perip heral position, both topic construc - tions of v erbs and dislo cation cop ying of v erbs r equir e no marking, but the latter in v ol v e a v erb in the right perip hery . M orp ho-syntactic marking is optional in ‘ ev en ’ - f ocus constructions of v erbs (i.e., lin ), but it is obligatory in copula f ocus constructions of v erb (i.e., hai ). This illustrates that the f our v erb doubling constructions ar e distinctiv e in morp ho-syntactic term. As will be discussed shortl y , they ar e also distinguishable in terms of discourse eff ects. 3.3.3 Discour se eff ects As far as interpr etation is concerned, the f our v erb doubling constructions ar e not discourse- neutral, w hen compar ed to the non-doubled counterparts. They ar e generall y incompatible with out -of -the- 58 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS blue contexts or the beginning of a story . This subsection describes the diff er ent discourse eff ects associated with these constructions in terms of inf ormation structur e. 3.3.3.1 Contr asti v e v erbal topi cs In addition to topicality , it is suggested that topic constructions of v erbs ar e used “to p lace tw o v erbs in contrast” (M atthew s and Y ip 2011 , p.88). I t is natural to ha v e a continuation that contrasts the v erb in the pr evious sentence. (71) Contrasting tw o v erbs in topic constructions of v erbs M aai buy k eoi s/he hai co p maai - zo buy -perf ni -bun this-cl syu. book Batgw o but mei not. y et tai. r ead ‘ As f or buying, s/he has bought this book. But ( s/he ) ha v e r ead (it ). ’ If a v erbal alternativ e is contextuall y una v ailable or pragmaticall y odd, a contrastiv e r eading in topic constructions of v erbs is difficult to obtain, r esulting in inf elicity . 27 This can be illustrated with the copula v erb hai or the v erb sing ‘ha v e.surname’ . (72) V erbs that fail in topic constructions of v erbs a. # H ai , co p Aaming Aaming hai co p daaihoksaang univ ersity .student ‘ Aaming is a univ ersity student. ’ b. # S ing , ha v e.surname Aaming Aaming hai co p sing ha v e.surname Lei Lee ‘ Aaming has “Lee” as his surname. ’ This suggests that the doubled v erb serv es as a contrastiv e topic in the construction. Cheng and V icente ( 2013 , p.5, and fn.5) note that this construction usuall y carries verum f ocus , w hich affirms the 27. I thank Ka- F ai Y ip f or discussions. 59 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS truth of the pr oposition. This can be attributed to the pr esence of the copula hai bef or e the v erb. Combining these ideas, the interpr etation of ( 71 ) is that (i) the action of buying is contrasted with r eading, and (ii) the sentence f ocuses on the truth of the pr oposition (i.e., ‘ s/he has bought this book’). I discuss a piece of evidence fr om the or dering with r egar d to topics in support of the contrastiv e topic status of the doubl ed v erb. The sentences in ( 73 ) sho w that the doubl ed v erb has to f ollo w a ( n on-contrasting) discourse topic. 28 (73) A topic must pr ecede a contrastiv e ( v erbal) topic a. Gaauzi dump ling ne1, to p sik , eat ngo I hai co p sik eat ge2. sfp ‘Dump lings, as f or eating, I eat ( them ). ’ b. ?? S ik , eat gaauzi dump ling ne1, to p ngo I hai co p sik eat ge2. sfp ‘ As f or eating, dump lings, I eat ( them ). ’ This f ollo w s if contrastiv e topics ar e indeed a sub-type of f ocus ( topic f oci, f ollo wing Büring ( 1997 ) and Krifka ( 2008 )), and if genuine topics ha v e to pr ecede f oci in the CP perip hery in Cantonese, i.e., the topic field is higher than the f ocus field ( as argued f or in C. C.-H. Cheung 2015 ). 3.3.3.2 A dditi v e v erbal f oci I t is suggested that the ‘ ev en ’ - f ocus constructions in M andarin con v ey an addictiv e f ocus r eading with scalar interpr etation, similar to English even (Sh yu 1995 ; Badan 2007 ). This idea can be app lied to ‘ ev en ’ - f ocus constructions of v erbs. F or examp le, in ( 74 ), lin picks out an ev entuality , i.e., mong ‘look’ , w hich r epr esents one of the extr emes on the scale of things that Aaming does not dar e to do. The speak er asserts that this ev entuality holds true among the things that Aaming does not dar e to do to Aafan. I t imp lies that all other non-extr eme ev entualities on the same scale, f or examp le, talking to 28. S imilar examp les ar e discussed briefl y in Cheng and V icente ( 2013 , p.5, fn.4). 60 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS Aafan, ar e also true. (74) A ddictiv e f ocus in ‘ ev en ’ - f ocus constructions of v erbs Lin ev en mong look Aaming Aaming dou also m-gam not -dar e mong look Aafan. Aafan M - hou not -good w aa sa y tung with k eoi her gongje. talk ‘ Aaming doesn ’t ev en dar e t o LOO K A T Aafan, not to mention to talk to her . ’ In the discussions of M andarin cases, Cheng and V icente ( 2013 ) observ es that man y cases of ‘ ev en ’ - f ocus constructions of v erbs tend to occur with negation, w hich is not observ ed with the p hrasal coun- terparts. They suggest that this is due to the scalarity r equir ement imposed on v erbs, as it r equir es the v erb to be connected to a scale. A scale can be established b y polatrity ( e.g., negation ), or superlativ es, as illustrated with the Cantonese in ( 75 ). (75) Scalarity satisf ed b y a superlativ e expr ession lin ev en jam drink Aaming Aaming dou also jiu w ant jam drink zeoi most peng cheap ge m o d jejam. drink ‘ Aaming ev en w ants to D RINK f the cheapest drink. ’ H o w ev er , ther e ar e cases of ‘ ev en ’ - f ocus construction of v erbs that in v ol v e neither negation or superlativ es. M atthew s and Y ip ( 1998 ) notes that a v erb suffix ed b y a potential particle -dou can also license ‘ ev en ’ - f ocus constructions of v erbs. (76) V erbs with a potential suffix (M atthew s and Y ip 1998 , p.183) H aang w alk dou also haang -dou w alk -able laa1. sfp Gam so kan! close ‘W e can ev en W ALK ther e. So close!’ A dditionall y , r ep lacing the superlativ e in the sentence in ( 75 ) with a definite expr ession is indeed pos- sible. 29 The v erb in ( 77 ) is not associated with negation or potential suffix either . I ther ef or e suggest 29. I thank Ka- F ai Y ip f or pointing out this to me. 61 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS that ther e is no specific r equir ement on ‘ ev en ’ - f ocus constructions of v erbs. (77) V erbs with no marking (p.c. Ka- F ai Y ip ) Lin ev en jam drink Aaming Aaming dou also jiu w ant jam drink ni - zek this-cl paaizi. brand. M -daanzi N ot -onl y sik. eat. ‘ Aaming ev en w ants to D RINK (pr oducts of ) this brand, not just eating (pr oducts of this brand). ’ W ithout going into the pr ecise semantic f ormulation of ‘ ev en ’ - f ocus constructions, I f ollo w Sh yu ( 1995 ) and Badan ( 2007 ) and assume that the ‘ ev en ’ - f ocus constructions con v ey both additivity and scalarity . P articularl y , the addivitiy can be attributed to the f ocus particle lin (Cheng and V icente 2013 ), w her eas the scalarity deriv es fr om the maximality operator dou , w hich pr o vides a scalar ex - tr eme (Giannakidou and Cheng 2006 ; Xiang 2008 ; Sh yu 2016 ). S imilar to v erbal t opics, the v erbs mark ed b y lin ha v e to f ollo w discourse topics ( cf. Badan 2007 ), further confirming their f ocus natur e. (78) A topic must pr ecede the lin - mark ed v erb a. N i -bou this-cl dinnou computer lin ev en jung use Aaming Aaming dou also mou not.ha v e jung -gw o use-exp ‘This computer , Aaming didn ’t ev en USE (i t ). ’ b. * Lin ev en jung use ni -bou this-cl dinnou computer Aaming Aaming dou also mou not.ha v e jung -gw o use-exp Int.: ‘This computer , Aaming didn ’t ev en USE (it ). ’ 3.3.3.3 Exhausti v e v erbal f oci The signatur e f eatur e of copula f ocus constructions is the exhaustiv e identification function (P aris 1998 ; C. C.-H. Cheung 2015 ; P an 2019 ). Descriptiv el y , w hen a v erb is targeted in the construction, an exhaustiv e set of ev entualities is pr esupposed. The sentence asserts that a giv en pr oposition holds 62 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS true of the ev entuality denoted b y the v erb, and all other alternativ es in the set ar e false (C. C.-H. Cheung 2015 ; Z ubizarr eta and V ergnaud 2017 , among others ). This idea can be illustrated with the conjunction tests, w hich diagnose exhaustivity (C. C.-H. Cheung 2015 ; Z ubizarr eta and V ergnaud 2017 ). In ( 79 ), the first sentence indicates that the ev entuality of touching is the only one that holds true among things that Aaming does not dar e to do to the animal. I t is incompatible with the f ollo w -up clause that asserts another ev entuality (i.e., ‘look’) that also holds true. 30 (79) Conjunction tests as a diagnostic f or exhaustivity H ai co p dim touch Aaming Aaming m-gam not -dar e dim touch ni - zek this-cl dungmat animal ze1. sfp *K eoi he zung also m-gam not -dar e mong look tim1. sfp ‘ Aaming dar e not to T O U CH this ani mal onl y . H e also dar e not to look at (it ). ’ P a n ( 2019 ) notes that ther e is a r estriction on episodicity in copula f ocus constructions in M an- darin, w hich pr ecludes episodic ev entualities in copula f ocus constructions. H o w ev er , no such r e- striction is observ ed with copula construction of v erbs in Cantonese. 31 (80 ) N o r estriction on ( non-)episodicity H ai co p fan sleep Aaming Aaming fan - zo sleep-perf singjat w hole.da y ze1. sfp M - hai not -co p w aan- zo p la y -perf singjat. w hole.da y ‘ Aaming onl y SLEPT the w hole da y . H e didn ’t hang out the w hole da y . ’ 30. The exhaustiv e identification ma y be attributed to the copula hai or to the SFP ze1 . One possibility is that the SFP ze1 supp lies the exhaustiv e semantics (F ung 2000 ), and the copula hai r estricts the f ocus scope to the immediatel y f ollo wing constituent, e.g., the v erb dim in ( 79 ). The choice does not bear on the discussion, ho w ev er . 31. In fact, the r estriction does not seem to hold in Cantonese in general. (i) N o r estriction on ( non-)episodicity in copula f ocus construction of objects H ai co p ni -bou this- cl dinnou computer Aaming Aaming camjat y ester da y zing - laan- zo mak e-br ok en- perf ze1. sfp K eita other dinnou computuer dou all mou not.ha v e mantai. pr oblem ‘I t is onl y this computer that Aaming has br ok en y ester da y . Other computers ar e fine. ’ 63 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS As a final r emark, or dering r estrictions with r espect to discourse topics sho w that hai - mark ed v erbs ar e similar to v erbal topics and lin - m ark ed v erbs: they must f ollo w the discourse topic. (81) A discourse topic must pr ecede the hai - mark ed v erb a. T ungseon fluent ge m o d jatman J apanese hai co p gong speak Aaming Aaming gong - m-dou speak - not -able ze1. sfp ‘ Aaming cannot SP EAK fluent J apanese onl y . ’ b. * H ai co p gong , speak tungseon fluent ge m o d jatman, J apanese Aaming Aaming gong - m-dou speak - not -able ze1. sfp ‘ Aaming cannot SP EAK fluent J apanese onl y . ’ 3.3.3.4 Def ocused v erbs The discourse eff ects br ought along b y dislocation cop ying, as suggested in L. Y .-L. Cheung ( 2015 ), concern “ emp hasis, clarification and r epair” , but he also notes that “these pragmatic functions ar e r elativ el y w eak” (p.262). The pr ecise inf ormation structural status of the doubled v erbs, or mor e gen- erall y , the doubled elements, r emains unclear in the literatur e. Ther e ar e tw o appar entl y conflicting view s. On one hand, L. Y .-L. Cheung ( 2015 ) and K. K. Chan ( 2016 ) suggest that the doubled elements r eceiv e minor emp hatic interpr etation ( or a special kind of contrastiv e interpr etation ). On the other hand, Lee ( 2017 , 2020 ) suggests that the post -sentence- final particle position is designated f or def o- cused elements. Lik ewise, Lai ( 2019 ) mentions in passing that the dislocated elements ar e mark ed as giv en. In w hat f ollo w s, I first sho w that the dislocated elements ar e not topics or inf ormational f ocus. I then suggest ho w the tw o view s ar e indeed compatible with each other . P articularl y , I suggest that the doubled elements ar e def ocused, and the r eported emp hatic eff ects come fr om the pragmatic eff ects of r epetition of expr essions. R ecall the observ ation in § 3.3.1 and § 3.3.3.1 that dislocation cop ying of v erbs can target modal 64 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS and the copula v erb, w hich cannot be topicalized. This suggests that the v erb in dislocation cop ying is qualitativ el y diff er ent fr om that in topic constructions, speaking against their status as v erbal topics. F urthermor e, adopting the question-answ er pair test in L. Y .-L. Cheung ( 1997 ) and K. K. Chan ( 2016 ), the question-answ er pairs in ( 82 ) and ( 83 ) sho w that the doubled v erb cannot serv e as the answ er to a question that f ocuses on the v erb in as ( 82 a ), w her eas the same sentence can f elicitousl y answ er a question that f ocuses, f or examp le, the object, as in ( 83 a ). This suggests that the doubled v erbs cannot serv e as inf ormational f ocus either . (82) V erbs cannot serv e as answ ers in dislocation cop ying of v erbs a. N ei y ou gammaan tonight fan- m- fan sleep- not -sleep ni - zoeng this-cl cong bed aa3? sfp ‘W ill y ou sleep on this bed tonight?’ b. # N go I gammaan tonight fan sleep ni - zoeng this-cl cong bed aa3 sfp fan . sleep ‘I ( will) sleep on this bed tonight. ’ (83) Objects can serv e as answ ers in dislocation cop ying of v erbs a. N ei y ou gammaan tonight fan sleep bin- zoeng w hich-cl cong bed aa3? sfp ‘W hich bed will y ou sleep on tonight? ’ b. = ( 82 b ) N go I gammaan tonight fan sleep ni - zoeng this-cl cong bed aa3 sfp fan . sleep ‘I ( will) sleep on this bed tonight. ’ F urthermor e, the doubled elements ar e unlik el y to be interpr eted as contrastiv e f ocus, at least in the standar d sense, since they r esist contrastiv e str ess, as observ ed in L. Y .-L. Cheung ( 2015 , p.261). I illustrate this point with dislocation cop ying of v erbs ( w her e str ess is indicated with capitals ). 65 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS (84) N o str ess of the doubled v erb # N go I gammaan tonight fan sleep ni - zoeng this-cl cong bed aa3 sfp F AN . sleep ‘I ( will) sleep on this b ed tognight. ’ In light of these observ ations, L. Y .-L. Cheung ( 2015 ) exp lor es the possibility that the doubled elements mark a minor emp hasis, or a special type of contrastiv e f ocus ( that r esists contrastiv e str ess ), w hose function is “to highlight to the hear er a discr epancy betw een the assertion of the host clause and the speak er’ s supposition that the hear er is unlik el y to tak e the assertion into the common gr ound” (p.263). K. K. Chan ( 2016 ) similarl y suggests that it is the back gr ound assumptions, instead of the f ocus, that is contrasted in dislocation cop ying. Instead of positing a fine-grained distinction on the natur e of contrastiv e f ocus, I suggest that the position f ollo wing the sentence- final particles is designated f or hosting def ocused/ giv en elements (Lee 2017 , 2020 ; Lai 2019 ). The primary function of def ocusing an element is to “ enable a con- stituent to escape the f ocus domain and to r ealize its discourse linking in f ormal terms” (M olnárfi 2002 , p.1132). I suggest that def ocus/ antif ocus is the counterpart notion of f ocus and can be mark ed grammaticall y (f or discussions and cr oss- linguistic evidence, see Lee 2020 ). 32 This not onl y exp lains w h y the doubled elements r esist f ocus ( and topic ) interpr etation, but also unif orml y accounts f or the discourse functions of dislocation cop ying and right dislocation, w hich can be r egar ded as instantia- tions of a single pr ocess of def ocalization. As f or the observ ed emp hatic eff ects, I suggest that it is a pragmatic consequence of the r epeti - tion of expr essions in general: w hen the same expr ession appear mor e than once, it naturall y giv es rise to a sense of emp hasis, but it need not be pr ominent or contrastiv el y f ocused. 33 This is w h y the doubled elements do not pattern with contrastiv e f ocus ( as they r esist f ocus str ess ). Indeed, the mi - 32. I t is not uncommon that def ocused elements end up in the sentence- final position. See, f or examp le, T akano ( 2014 ) f or discussions in J apanese. 33. This possibility is also mentioned in passing in Lai ( 2019 ). Indeed, M artins ( 2007 ) discusses similar cases of v erb doubling in E ur opean P ortuguese and suggests that the constructions con v ey emphatic affirmation . 66 3.3. VERB DO UB LIN G CO NSTR U CTI O NS AND D ISCO URSE EFFECTS nor emp hatic eff ects as depicted in L. Y .-L. Cheung ( 2015 ) ar e general enough to app l y to other v erb doubling constructions as w ell, all of w hich disp la y s some lev el of emp hatic eff ects on the v erb. I ther ef or e conclude that the v erb in dislocation cop ying of v erbs is def ocused, a discourse function that is diff er ent fr om the other thr ee v erb doubling constructions. Lastl y , concerning the r elativ e or der betw een a def ocused v erb and a discourse topic, w hile a dis- course topic can appear in the right perip hery as in ( 85 a ), it cannot pr ecede or f ollo w a def ocused v erb as in ( 85 b ) and ( 85 c ). 34 (85) A discourse topic cannot pr ecede the def ocused v erb a. ngo I soeng w ant maai buy gaa3, sfp ni -bun this-cl syu book ‘I w ant to buy this book. ’ b. * ngo I soeng w ant maai buy gaa3, sfp ni -bun this-cl syu, book soeng w ant ‘I w ant to buy this book. ’ c. ?? ngo I soeng w ant maai buy gaa3, sfp soeng , w ant ni -bun this-cl syu book ‘I w ant to buy this book. ’ 3.3.3.5 Interim summary T aking stock, w hile the f our v erb doubling constructions con v ey diff er ent discourse eff ects, con- trastiv eness is a common pr operty among the first thr ee constructions. This is corr elated with the or dering r estriction with discourse topics, i.e., contrastiv e elements must f ollo w discourse topics. On the other hand, v erbs in dislocation cop ying disp la y an opposite pattern, w her e they r eceiv e a non- 34. The unacceptability in ( 85 c ) is r elativ el y less sev er e, especiall y w hen a pause is inserted betw een the v erb and the topic, but a pause does not impr o v e the sentence in ( 85 b ). 67 3.4. E VID EN CE FO R VERB M O VEMENT contrastiv e interpr etation and they cannot f ollo w a discourse topic. The observ ations in this subsec - tion ar e summarized in T able 3.7 . T ype Discourse eff ects Contrastiv eness R elativ e or der with topics T opic - V contrastiv e topic 4 T opic > V / *V > T opic ‘E v en ’ - V additiv e f ocus 4 T opic > V / *V > T opic Copula- V exhaustiv e f ocus 4 T opic > V / *V > T opic DC- V def ocused/ giv en elements 6 *T opic > V / ??V > T opic T able 3.7: The discourse eff ects of the v erb doubling constructions I str ess that these discourse eff ects observ ed in v erb doubling constructions ar e largel y similar , if not identical, to their p hrasal counterparts in ( 27 ), as described in the literatur e. The onl y diff er ence is the types of elements (i.e., heads v s. p hrases ) w hich ar e targeted. 3.4 E vi dence f or v erb mo v ement In this section, I pr esent evidence in fa v or of a mo v ement anal y sis of v erb doubling constructions. I suggest that the v erb in the perip heral position and the v erb in the base position ar e deriv ationall y r elated b y mo v ement dependencies. E vidence comes fr om (i) lexical identity eff ects, (ii) island eff ects, and (iii) idiomatic expr essions. 3.4.1 Le xi cal i dentity eff ects This subsection r eports the lexical identity eff ects in v erb doubling constructions. P articularl y , the v erb in the (left or right ) perip hery must be identical to the v erb in the base position. I argue that such identity eff ects r ev eal a dependency r elation betw een the tw o v erbs. M or e specificall y , the depen- dency r elation is syntactic , rather than semantic , i.e., tw o v erbs must be the same lexical item, instead of sharing identical meaning, or standing in an entailment r elation. F ollo wing much w ork in v erb doubling/ clefting (Cable 2004 ; V icente 2007 ; Cheng and V icente 2013 ), I argue that lexical identity 68 3.4. E VID EN CE FO R VERB M O VEMENT eff ects ar e a dir ect consequence of syntactic mo v ement. I t f ollo w s straightf orw ar dl y fr om the cop y theory of mo v ement (Chomsky 1995b , et seq. ), w her e syntactic mo v ement cr eates a chain of multip le copies of the mo ving elements in the structur e. 35 In or der to confirm the identity eff ects ar e b y natur e syntactic, instead of semantic, it is instructiv e to see w hether non- identical v erbs can appear in v erb doubling constructions. Especiall y in topic constructions, tw o elements ma y disp la y the so-called genus -species eff ects , w her e an element stands in an asymmetric entailment r elation with another . F or examp le, a base-generated (frame-setting) topic bears such a r elation with a ( mor e specified) nominal in the sentence, as in ( 86 ), w her e tuna entails fish . (86) N o l exical identity eff ects with base generated topics J yu fish ngo I zungji lik e sik eat tanlaa. tuna ‘ As f or fish, I lik e eating t una. ’ If v erb doubling constructions disp la y a similar pattern, it can be tak en as a piece of evidence against a mo v ement dependency betw een the tw o v erbs ( cf. Cable 2004 ). I emp lo y f our pairs of v erbs that ar e in genus-species r elation, listed in ( 87 ). (87) P airs of v erbs in asymmetric entailment r elations a. caau ‘to fry’ entails zyu ‘to cook’ b. paau ‘to run ’ entails juk ‘to mo v e’ c. mo ‘to pet’ entails dim ‘to touch ’ d. f ei ‘to fl y t o ’ entails heoi ‘to go to ’ App l ying these v erbs to the v erb doubling constructions, it is observ ed that the mor e general v erb cannot appear in the perip heral position. This speaks against a base generation anal y sis. 35. The doubling eff ects ar e not commonl y observ ed with other instances of p hrasal mo v ement. I argue in Chapter 5 that this r esults fr om independent p honological r equir ements. 69 3.4. E VID EN CE FO R VERB M O VEMENT (88) Lexical identity , but not semantic entailment, is crucial a. f caau / fry *zyug cook ngo I hai co p soeng w ant caau fry coi v egetable ge2. sfp ‘ As f or frying/ cooking, I w ant to fry the v egetables. ’ b. Lin ev en f paau / run *jukg mo v e Aaming Aaming dou also m-gam not -dar e paau . run ‘ Aaming doesn ’t ev en dar e to R UN/M O VE. ’ c. H ai co p f mo / pet * dimg touch Aaming Aaming m-gam not -dar e mo pet ni - zek this-cl dungmat animal ze1. sfp ‘ Aaming dar e not to P ET /T O U CH this animal onl y . ’ d. Aaming Aaming haanin next. y ear f ei fl y M eigw ok US aa3 sfp f f ei / fl y / *heoig go ‘ Aaming ( will) fl y to US next y ear . ’ N ote that no lexical identity eff ects ar e observ ed with sentences with zau ‘then ’ . As suggested in § 3.3.2.1 , these sentences ma y in v ol v e a bi -clausal structur e with the first clause being a r educed conditional minimall y containing a v erb. This suggests the absence of syntactic dependencies betw een the tw o v erbs. (89) N o identity eff ects in sentences with zau ‘then ’ a. Context: the interlocutors ar e discu ssing whether to cook at home or dine out. H ai co p zyu cook ngo I zau then jigaa no w caau fry coi. v egetable ‘(If it is decided to ) cook ( at home ), then I fry the v egetables no w . ’ b. Context: the interlocutors ar e discu ssing whether to go to the cinema. 70 3.4. E VID EN CE FO R VERB M O VEMENT H ai co p heoi go ngo I zau then ceotfat depart laa3. sfp ‘(If it is decided to ) go, then I depart no w . ’ On the other hand, it is also instructiv e to consider another possible semantic dimension of the identity eff ects, namel y , semantic identity . H er e I adopt pairs of v erbs that ar e ( nearl y ) semanticall y identical to each other . Thr ee of them in v ol v e English loan w or ds w hich ha v e been p honologicall y adapted to the p honotactics in Cantonese. F or examp le, cek1 is f or ‘ check’ , kip1 f or ‘k eep ’ and pi6sen1 f or ‘pr esent’ ( with tones indicated b y the number ). Another pair concerns tw o v erbs of selling, namel y , f ong ‘let.go ’ and maai ‘ sell’ . They ar e semanticall y identical in the context of, f or examp le, stock mark ets (but not all other contexts ). App l ying these v erbs to v erb doubling constructions, it is observ ed that the v erbs in the perip hery ha v e to be lexically identical to the v erb in the base position. Their semanticall y identical counterparts ar e not acceptable in the same position. (90 ) Lexical identity , but not semantic identity , is crucial a. f caa / check/ * cekg check ngo I hai co p caa -gw o check -exp ni -go this-cl jan person ‘ As f or checking, I ha v e check ed this person. ’ b. Lin ev en f kip / k eep *bougung k eep Aaming Aaming dou also m-soeng not - w ant kip . k eep ‘ Aaming dones’t ev en w ant to KEEP (it ). ’ c. H ai co p f pisen / pr esent *bougoug pr esent k eoi s/he mou not.ha v e seonsam confidence pisen -dak pr esent -res hou good ze1. sfp ‘S/he lacks the confidence to P RESENT w ell onl y . ’ 71 3.4. E VID EN CE FO R VERB M O VEMENT d. Aaming Aaming tingjat tomorr o w wui will f ong see ni -di this-ccl gupiu stock aa3 sfp f f ong / sell *maaig sell ‘ Aaming will sell these stocks tomorr o w . ’ H o w ev er , in sentences with zau ‘then ’ , the lexical identity eff ects disappear . (91) N o identity eff ects in sentences with zau ‘then ’ a. H ai co p cek check ngo I zau then tingjat tomorr o w faan- heoi back -go caa- haa check -d el ngo m y bun cl geisibou. notebook ‘(If it is decided to ) check, then I go back and check m y no tebook tomorr o w . ’ b. H ai co p kip k eep ngo I zau then jigaa no w bougun- zyu k eep-co nt sin1. sfp ‘(If it is decided to ) k eep, then I k eep it f or no w first. ’ I conclude that the lexical identity eff ects in v erb doubling constructions r ev eal syntactic depen- dencies, rather than semantic dependencies, betw een the tw o v erbs. U nder the cop y theory of mo v e- ment, the doubling eff ects ar e a natural consequence of mo v ement, w her e multip le copies ma y be p honologicall y r ealized in a mo v ement chain. I tak e this as evidence f or a mo v ement anal y sis of v erb doubling constructions. Bef or e I lea v e the discussions of lexical identity eff ects, ther e is an intriguing but puzzling diff er - ence betw een v erb doubling constructions. V erbs in topic constructions, ‘ ev en ’ - f ocus constructions and copula f ocus constructions onl y allo w doubling of the v erb in bar e f orm, to the ex clusion of v erbal suffix es. H o w ev er , the v erb in dislocation cop ying r equir es doubling of both the v erb and its associ - ated suffix es. The diff er ence is illustrated in the sentences in ( 92 ), w hich contrast topic constructions of v erbs and dislocation cop ying of v erbs. 72 3.4. E VID EN CE FO R VERB M O VEMENT (92) An asymmetry on lexical identity eff ects among v erb doubling constructions a. N go -(* dou ) hungry -res Aaming Aaming hai co p ngo -dou hungry -res w an dizzy laa3. sfp ‘ As f or being hungry , Aaming is so hungry that he f eels dizzy . ’ b. ( cf. L. Y .-L. Cheung 2015 , p.229) Aaming Aaming ngo-dou hungry -res w an dizzt laa3 sfp ngo- *( dou ). hungry -res ‘ Aaming is so hungry that he f eels dizzy . ’ One possible exp lanation to this asymmetry is to connect the diff er ence to the diff er ent discourse eff ects of these constructions. F or examp le, it ma y be that v erbs in contrastiv e interpr etations can onl y r epr esent the ev entuality pr oper , and thus onl y the bar e f orm of the v erb is targeted f or interpr etation. On the other hand, in non-contrastiv e (i.e., def ocused) interpr etation, the r estriction does not app l y and t hus a suffix ed v erb can be targeted as a w hole. 36 37 3.4.2 I sland eff ects This subsection examines the possible structural distance betw een the v erb in the perip heral position and the v erb in the canonical position. S ince R oss ( 1967 ), it has generall y been agr eed that certain structural domains ar e inaccessible f or syntactic operations or dependencies. In this subsection, I sho w that the v erbs in v erb doubling constructions ar e sensitiv e to “island eff ects, ” a typical charac - teristic of mo v ement dependencies. F urthermor e, I sho w that the tw o v erbs can occur at a distance and tolerate a clausal ( CP ) boundaries. 36. This, ho w ev er , does not exp lain w h y doubling of the suffix is obligatory , instead of optional, in dislocation cop ying. I ha v e to lea v e this issue open. 37. Another possibility is to connect the asymmetry to the sp lit betw een left and right perip hery of the v erb doubling constructions. I lea v e this possibility f or futur e r esear ch. 73 3.4. E VID EN CE FO R VERB M O VEMENT 3.4.2.1 I sland sensiti vity The v erb in the perip heral position and the v erb in the base position cannot be interv ened b y “island” boundaries. T ypical “islands” include (i) comp lements of a noun p hrase, (ii) r elativ e clauses, (iii) adjunct clauses and (iv ) sentential subject s. The first tw o can be subsumed under the S ubjacency Condition (Chomsky 1973 , 1981 ), w her eas the last tw o fall under the Condition on Extraction Domain (H uang 1982 ). The f ollo wing f our sets of data sho w that the v erb doubling constructions become unacceptable if the base v erb originates in these domains, r espectiv el y illustrated in ( 93 ), ( 94 ), ( 95 ), and ( 96 ). (93) NP comp lement a. * T ai r ead ngo I tungji agr ee go-go that -cl [k eoi s/he hai co p tai -gw o ] r ead-exp ge m o d jigin. opinion Int.: ‘ As f or r eading, I agr ee with the opinion that s/he has r ead (it ). ’ b. * Lin ev en zau lea v e ngo I gamjat toda y dou also ting dou hear [ Aaming Aaming zau -dak lea v e-res maan-gw o slo w -than jan ] person ge m o d siusik. rumor Int.: ‘T oda y I hear that rumor that Aaming is slo w er than others ev en in LEA VIN G. ’ (94) R elativ e clauses a. * H ai co p dim touch ngo I jicin once gin-gw o see-exp ni - zek this-cl [ Aaming Aaming m-gam not -dar e dim ] touch ge m o d dungmat. animal Int.: ‘I once sa w this animal - one that Aaming dar e not to T O U CH onl y . ’ b. * Aaming Aaming soeng w ant hok learn [ Aafan Aafan hoji can gong] speak ge m o d ni -saam- zung this-thr ee-cl jyujin language aa3 sfp hoji . can Int.: ‘ Aaming w ants to learn these thr ee languages that Aafan can speak. ’ 74 3.4. E VID EN CE FO R VERB M O VEMENT (95) A djuncts a. * tai r ead [ hai at Aaming Aaming hai co p tai - jyun r ead- finish bun cl syu book zihau ], after ngo I sin first faan-dou r eturn-arriv e ukk ei. home Int.: ‘ As f or r eading, I w as back after Aaming has alr eady finished r eading the book. ’ b. * Lin ev en zau , lea v e [jan w ai because Aaming Aaming dou also zau-dak lea v e-res maan-gw o slo w -than jan ], person soji so ngo I mou not.ha v e dang w ait k eoi. him ‘S ince Aaming is slo w er than others ev en in LEA VIN G, I didn ’t w ait f or him. ’ (96) Sentential subjects 38 a. * H ai co p gong speak [ Aaming Aaming gong - m-dou speak - not -able jatman ] J apanese ling mak e Aafan Aafan hou v ery satmon. disappointed ‘That Aaming cannot SP EAK J apanese onl y disapp oints Aafan. ’ b. * [ Aaming Aaming gammaan tonight fan sleep ni - zoeng this-cl cong] bed jiging alr eady mouhobeimin una v oidable laa3 fan . ‘That Aaming ( will) sleep on this bed tonight is una v oidable. ’ S ince island sensitivity is typicall y tak en as evidence f or mo v ement dependencies, I suggest that these observ ations support a mo v ement anal y sis of v erb doubling constructions. 38. M atthew s and Y ip ( 1998 ) r eport that topic constructions of v erbs can occur acr oss sentential subject islands. They pr o vide the f ollo wing examp le, but m y inf ormants r eport that this sentence is unacceptable, contrary to their observ ation. (96) Sentential subjects and topic constructions of v erbs * (M atthew s and Y ip 1998 , p.181-182) T eoijau r etir e ngo I gok dak think 65 65 seoi y ear .old mei not. y et teoijau r etir e mou not.ha v e mantai. pr omblem ‘ As f or r etiring, I think not r etiring at 65 y ears old is fine. ’ 75 3.4. E VID EN CE FO R VERB M O VEMENT 3.4.2.2 Long -distance/ Cr oss-clausal dependenc i es W hile the tw o v erbs in v erb doubling constructions cannot be separated b y island boundaries, they can be interv ened b y a clausal/ CP boundary . In the sentences in ( 98 ), the clausal boundary is established b y attitude v erbs or v erbs of sa ying, so the clausal boundary is pr esumabl y as large as a CP . N ote that ( 98 d) in v ol v es multip le CP boundaries. (98) Cr oss-clausal dependencies in v erb doubling constructions a. Soeng w ant ngo I lam think [ Aaming Aaming hai co p soeng w ant heoi go ni -go this-cl wuiji ] meeting ge2. sfp ‘I think Aaming w ants to go to this meeting. ’ b. Lin ev en maai buy ngo I gu guess [ Aaming Aaming dou also mei not. y et maai buy ni -bun this-cl syu.] book ‘I guess Aaming has not ev en BO U GHT this book. ’ c. H ai co p zou r ent ngo I gok dak think [ Aaming Aaming m-soeng not - w ant zou r ent san new lau ] house ze1. sfp ‘I think Aaming does not w ant to RENT a new ho use onl y . ’ d. N go I ting hear [ tinhei w eather bougou r eport w aa sa y [ tingjat tomorr o w honang possible lokjyu rain w o5] sfp ] honang possible ‘I hear d fr om the w eather r eport t hat it ma y rain tomorr o w . ’ Although long -distance/ cr oss-clausal dependencies do not necessitate a mo v ement anal y sis 39 , they ar e pr edicted b y a mo v ement anal y sis to be possible. N ote that these case demonstrate the HM C- violating pr operty of v erb doubling constructions, since, the dependencies betw een the tw o v erbs ar e interv ened b y o v ert heads, namel y , the embedding attitude v erbs or the v erbs of sa ying. 39. This is because, f or examp le, an unselectiv e binding appr oach can also captur e long -distance dependencies (H eim 1982 ). F or pr oposal and app lication on interr ogativ e clauses, see T sai ( 1994 ), among others. 76 3.4. E VID EN CE FO R VERB M O VEMENT 3.4.2.3 Interim summary The locality eff ects with r egar d to (im )penetrability can be summarized in ( 99 ). The dependencies be- tw een the tw o v erbs in v erb doubling constructions hold acr oss CP boundaries, but not island bound- aries. (99) A schematic r epr esentation of the locality eff ects in v erb doubling constructions V perip hery ... 8 > < > : CP boundaries *I sland boundaries 9 > = > ; ... V base 3.4.3 I di omati c e xpr essi ons Separability of idiomatic expr essions, is often tak en as evidence f or mo v ement. The v alidity of this ar - gument builds on a particular assumption on the anal y sis of idiomatic expr essions in natural language. I t is commonl y assumed that an idiomatic expr ession must f orm a constituent, either in the lexicon (J ack endoff 1997 ) or in a local domain in the course of deriv ation (M arantz 1997 ). W ith either one of these assumptions, idiom chunks can be used to diagnose the base position of the disp laced elements. In our cases of v erb doubling, a mo v ement anal y sis pr edicts that the v erb in an idiomatic expr ession might be able to occur in v erb doubling constructions without losing its idiomatic meaning. 40 This is borne out in ( 100 ). (100 ) I diomatic expr essions in v erb doubling constructions a. Put pour ngo I hai co p honang possible put - zo spill -perf nei y ou laang cold seoi. w ater Batgw o but ngo I zihai onl y soeng w ant giklai encourage nei y ou ze1. sfp I diomatic r eading: ‘I ma y ha v e spoiled y our p leasur e, but I onl y w ant to encourage y ou. ’ 40. The same test is adopted to diagnose mo v ement f or copula f ocus constructions in C. C.-H. Cheung ( 2015 ). 77 3.4. E VID EN CE FO R VERB M O VEMENT b. Lin ev en ceoi blo w Aaming Aaming dou also m-tung not - with ngo me ceoi blo w seoi. w ater M - hou not -good w aa sa y king discuss zingging serious je. thing I diomatic r eading: ‘ Aaming didn ’t ev en CHIT -CHA T with me, not to mention discuss serious t hings. ’ c. H ai co p duk pok e Aaming Aaming duk -gw o pok e nei y our buizek back ze1. sfp K eoi he mou not.ha v e dongmin face.to.face naau -gw o scold-exp nei y ou aa3. sfp I diomatic r eading: ‘ Aaming h as ST AB B ED A T Y O UR B A CK onl y . H e hasn ’t scold at y ou face to f ace. ’ d. N i - zek this-cl laaihaamou toad soeng w ant sik eat tinngo sw an juk meat aa3 sfp soeng . w ant I diomatic r eading: ‘This person is cra ving f or w hat s/he is not w orth y of. ’ I t should be noted that in ( 100 a-c ), the contrastiv e interpr etation does not fall ex clusiv el y on the v erb in the perip heral position. Instead, it is the w hole VP that is f ocused. This can be illustrated b y the continuing clauses that contrast the first clause with another VP . This observ ation is b y no means unique to idiomatic expr essions, as illustrated with the ‘ ev en ’ - f ocus construction of v erbs belo w . (101) VP f ocus interpr etation in ‘ ev en ’ - f ocus constructions of v erbs Lin ev en sik eat jau ha v e jan person dou also sik - m-saai eat - not - finish ni - fan this-cl saaleot. salad M - hou not -good w aa sa y jam- maai drink -also bui cl gaaf e. coff ee ‘Ther e is someone w ho cannot ev en FINISH THIS SALAD , not to mention drink the coff ee. ’ 78 3.5. FOCUS INTER VENTI O N EFFECTS The mor e salient VP - f ocus r eading (i.e., mor e salient than a V - f ocus r eading) is pr esumabl y due to the fact that the idiomat ic expr essions ar e VP (V+O ) expr essions and that it is difficult, if not impossible, to f ocus on the v erb to the ex clusion of the object ( w hile maintaining the idiomatic r eadings ). 41 3.5 F ocus Interv enti on Eff ects Bef or e I turn to a mo v ement anal y sis of the v erb doubling constructions in Cantonese, I further sho w that the dependency betw een the v erb in the base position and the v erb i n the perip heral position ma y be disrupted b y interv ention of an element with certain qualities. In w hat f ollo w s, I examine thr ee types of potential interv ening elements: v erbs, f ocused elements, and quantificational elements. The empirical observ ations with r egar d to interv ention eff ects ar e summarized in ( 102 ). (102) Interv ening elements in v erb doubling constructions a. V erbs/H eads do not cause interv ention eff ects. b. F ocused elements cr eate interv ention eff ects ( ex cept in dislocation cop ying of v erbs ). c. Quantificational elements do not cause interv ention eff ects. The findings r ev eal that the v erb doubling constructions exhibit interv ention eff ects trigger ed b y f ocused elements, but not elements of the same structural types (i.e., heads ). 3.5.1 N o interv enti on b y heads As discussed in § 3.2 , head mo v ement is said to be constrained b y the H ead M o v ement Constraint (T ra vis 1984 ) or R elativized Minimality (Rizzi 1990 ). This suggests that interv ention eff ects w ould be induced b y interv ening v erbs/heads in case of v erb mo v ement. H o w ev er , the f ollo wing data sho w that v erbs in the perip heral positions in v erb doubling constructions tolerate interv ening v erbs. In 41. The mismatch betw een morp ho-syntactic f ocus marking and the scope of f ocus interpr etation falls into another inter esting line of study concerning anti-pied-piping (B ranan and Erlewine 2020 ), w hich does not bear on the discussion her e, ho w ev er . 79 3.5. FOCUS INTER VENTI O N EFFECTS all the cases in ( 103 ), the v erbs in the base and perip heral positions ar e interv ened b y an ( o v ert ) head element, e.g., a contr ol v erb or a modal v erb. N ote that in terms of linear or der , the modal v erb wui in ( 103 d) does not “interv ene” the tw o copies of heoi ‘ go ’ . Anticipating a mo v ement anal y sis of dislocation cop ying, w her e the dislocated v erb is suggested to mo v e acr oss the modal v erb at some point in the deriv ation, the v erb in the perip heral position in ( 103 d) is comparable to the other thr ee constructions in terms of hier ar chy structur e . (103) N o interv ention eff ects trigger ed b y interv ening v erbs a. H eoi , go Aaming Aaming hai co p soeng w ant heoi go M eigw ok US ge2. sfp ‘ As f o r go ing, Aaming w ants to go to the US. ’ b. Lin ev en gong speak Aaming Aaming dou also m-sik not - kno w gong . speak ‘ Aaming doesn ’t ev en kno w ho w to SP EAK. ’ c. = ( 28 c ) H ai co p dim touch Aaming Aaming m-gam not -dar e dim touch ni - zek this-cl dungmat animal ze1. sfp ‘ Aaming d ar e not to T O U CH this animal onl y . ’ d. Aaming Aaming wui will heoi go maai buy ni -bun this- syu book aa3 sfp heoi . go ‘ Aaming will go to buy this book. ’ In other w or ds, v erb doubling constructions disp la y a HM C- violating pr operty . These cases ar e configurationall y unif orm ( at least at some point of their deriv ation ) with r espect to t he pr esence of interv ening heads. 80 3.5. FOCUS INTER VENTI O N EFFECTS (104) (The absence of ) interv ention eff ects in v erb doubling constructions, part 1 V perip hery ... 8 > > > > > > > > > > < > > > > > > > > > > : H ead soeng ‘w ant’ sik ‘ can ’ gam ‘ dar e’ wui ‘will’ 9 > > > > > > > > > > = > > > > > > > > > > ; V base ... 3.5.2 Interv enti on b y f ocused el ements W hile heads do not cause interv ention eff ects in v erb doubling constructions, this subsection r ev eals that f ocused elements disrupt the r elation betw een the v erb in the perip heral position and the v erb in the base position. T o illustrate this point, I adopt tw o types of (in-situ ) f o cused elements (i) f o- cused elements associated with dak ‘ onl y’ (T ang 2002 ), and (ii) wh - nominals, w hich bear inher ent f ocus (R ochemont 1986 ; H orv ath 1986 ; Shi 1994 ; S.-S. Kim 2006 ). As a r emark on the basic pr operties of dak ‘ onl y’ in Cantonese, dak ‘ onl y’ con v ey s an ( exhaustiv e ) f ocus r eading on the constituent to its right (T ang 2002 ). I t can appear in both pr ev erbal and postv erbal positions, ex emp lified in ( 105 ). (105) The distribution of the f ocus operator dak ‘ onl y’ a. Pr ev eal dak and s ubject f ocus (T ang 2002 , p.281) Dak onl y [ ngo ] F ocus I faatbiu expr ess jigin. opinion ‘Onl y I expr ess opinions. ’ 81 3.5. FOCUS INTER VENTI O N EFFECTS b. P ostv erbal dak and object f ocus (T ang 2002 , p.267) K e oi s/he tai -dak r ead-onl y [ saam-bun thr ee-cl syu ] F ocus book ‘S/he r ead onl y thr ee books. ’ In terms of syntactic category , T ang ( 2002 ) argues that the pr ev erbal dak is a v erb and the postv erbal dak is a v erbal suffix. H e also observ es that postv erbal dak imposes a car dinality r equir ement on its associate, w hich is absent in pr ev erbal dak . S ince the discussion her e f ocuses on the structural position of f ocused elements, these diff er ences do not bear on the argumentation. In w hat f ollo w s, I sho w that dak ‘ onl y’ ma y cause interv ention eff ects betw een the tw o copies of v erbs in thr ee out of f our v erb doubling constructions ( with the ex ception of dislocation cop ying of v erbs ). Specificall y , it causes interv ention eff ects w hen it is associated with a pr e- v erbal element ( e.g., a subject ) , but not with a post - v erbal element ( e.g., an object ). F irst, in the topic constructions of v erbs in ( 106 ), the dak -associate cannot appear in the subject position, as opposed to the object position. (106) S ubject v s. object f ocus in topic constructions of v erbs a. H eoi go dak onl y Aaming Aaming hai co p soeng soeng heoi heoi M eigw ok. US ‘ As f or going, onl y Aaming w ants to go to the US. ’ b. H eoi go Aaming Aaming hai co p heoi -dak go-onl y jat -go one-cl deif ong. p lace ‘ As f or going, Aaming has been to onl y one p lace. ’ N o te that the f ocus operator hai does not induce interv ention eff ects in ( 106 b ), suggesting that it is the f ocused elements, but not the f ocus particles/ operators that ar e r esponsible f or the interv ention eff ects. Indeed, it is possible to add another f ocus operator , zinghai ‘ onl y’ ( associating with the object ), bef or e the v erb in ( 106 b ), and no interv ention eff ects ar e observ ed ( the same is also observ ed in ( 108 b ) and ( 110 b )). 82 3.5. FOCUS INTER VENTI O N EFFECTS (107) F ocus operators do not trigger interv ention eff ects H eoi go Aaming Aaming hai co p jinghai onl y heoi -dak go-onl y jat -go one-cl deif ong. p lace ‘ As f or going, Aaming has been to onl y one p lace. ’ S imilarl y , in ‘ ev en ’ - f ocus constructions of v erbs, the dak -associate in the pr e- v erbal position leads to interv ention eff ects, as in ( 108 a ), as opposed to the dak -associate in the post - v erbal position, as in ( 108 b ). (108) S ubject v s. object f ocus in ‘ ev en ’ - f ocus constructions of v erbs a. * Lin ev en haang w alk dak onl y Aaming Aaming dou also haang - m-dou w alk - not -able sap ten fanzung. minute M - hou not -good w aa sa y paau. run ‘Onl y Aaming cannot ev en W ALK f or ten minutes, not to mention run. ’ b. Lin ev en haang w alk Aaming Aaming dou also jinghai onl y haang -dak w alk -able sap ten fanzung. minute M - hou not -good w aa sa y paau. run ‘ Aaming can W ALK onl y f or ten minutes, not to mention run. ’ One might suggest that the unacceptability of ( 108 a ) is due to the fact that the dak -associate in- terv enes the lin - V ... dou construction, instead of the V - V dependency . This is not the case, ho w ev er , since the interv ention eff ects persist ev en if dou occurs bef or e the dak -associate. This is sho wn in ( 109 ), w hich in v ol v es a bi -clausal structur e. 42 Cruciall y , dou surfaces in the matrix clause, w her eas the dak -associate is in the embedded clause. (109) Interv ention eff ects induced betw een the tw o v erbs, not lin and dou 42. F or discussions on the positions of dou , see Sh yu ( 1995 ) f or the M andarin counterpart lian ... dou constructions. 83 3.5. FOCUS INTER VENTI O N EFFECTS * Lin ev en haang w alk ngo I dou also jingw ai think [ dak onl y Aaming Aaming haang - m-dou w alk - not -able sap ten fanzung]. minute M - hou not -good w aa sa y paau. run ‘I think that onl y Aaming cannot ev en W ALK f or ten minutes, not to mention run. ’ In a parallel fashion, similar interv ention eff ects ar e observ ed in copula f ocus constructions of v erbs as in ( 110 ). (110 ) S ubject v s. object f ocus in copula f ocus constructions of v erbs a. * H ai co p maai buy dak onl y Aaming Aaming maai - m- hei buy - not -up ni -gaan this-cl uk house ze1. sfp K e oi he zou -dak - hei. r ent -able-up ‘Onl y Aamin g cannot B UY this house onl y . H e can ( aff or d) r enting it. ’ b. H ai co p maai buy Aaming Aaming jinghai onl y maai -dak buy -onl y jat -gaan one-cl uk house ze1. sfp Zou r ent gew aa if k eoi he hoji can zou r ent gei -gaan. sev eral -cl ‘ Aaming can B UY onl y one house onl y . If (it is about ) r enting, he can ( aff or d) r enting sev eral ho uses. ’ H o w ev er , it is important to note that similar interv ention eff ects do not r ep licate in dislocation cop ying of v erbs, as in ( 111 ): both sentences ar e acceptable. (111) S ubject v s. object f ocus in dislocation cop ying of v erbs a. dak onl y Aaming Aaming gammaan tonight hoji can fan sleep baat -go eight -cl zung hour zaa3 sfp hoji . can ‘Onl y Aaming can sleep f or eight hours tonight. ’ 84 3.5. FOCUS INTER VENTI O N EFFECTS b. Aaming Aaming gammaan tonight jinghai onl y hoji can fan-dak sleep-onl y jat -go one-cl zung hour zaa3 sfp hoji . can ‘ Aaming can sleep f or onl y one hour tonight. ’ I no w turn to wh -expr essions, w hich ma y induce similar interv ention eff ects as dak -associates. H er e, I f ollo w R ochemont ( 1986 ), H orv ath ( 1986 ), Shi ( 1994 ), and S.-S. Kim ( 2006 ) and assume that wh -expr essions bear inher ent f ocus ( despite the absence of o v ert morp ho-syntactic f ocus marking). W h -expr essions in the subject position ar e disallo w ed in v erb doubling constructions, but they ar e acceptable in the object position. This is illustrated with topic constructions of v erbs in ( 112 ) and ‘ ev en ’ - f ocus constructions v erbs in ( 113 ). (112) S ubject v s. object wh -expr essions in topic constructions of v erbs a. * Soeng w ant bingo w ho hai co p soeng w ant heoi go M eigw ok? US ‘W ho w ants to go to the US?’ b. Soeng w ant Aaming Aaming hai co p soeng w ant heoi go bindou? w her e ‘W her e does Aaming w ant to go?’ (113) S ubject v s. object wh -expr essions in ‘ ev en ’ - f ocus constructions of v erbs a. * Lin ev en haang w alk bingo w ho dou also haang - m-dou w alk - not -able sap ten fanzun? minute ‘W ho can ’t ev en W ALK f or ten minutes?’ b. Lin ev en haang w alk Aaming Aaming dou also haang - m-dou w alk - not -able geidou ho w .man y fanzung. minute ‘F or ho w man y minutes does Aaming ev en fail to W ALK?’ 85 3.5. FOCUS INTER VENTI O N EFFECTS T urning to copula f ocus constructions of v erbs, ther e is a comp lication. The sentences ar e unac - ceptable r egar dless of the position of the wh -expr essions, as in ( 114 a-b ). Indeed, in the absence of v erb doubling, the sentence in ( 114 c ) is still unacceptable. This suggests that copula f ocus constructions of v erbs ar e incompatible with wh -expr essions in general. So ther e is no evidence f or the interv ention eff ects in copula f ocus constructions fr om wh -expr essions. 43 (114) S ubject v s. object wh -expr essions in copula f ocus constructions of v erbs a. * H ai co p maai buy bingo w ho maai - m- hei buy - not -up ni -gaan this-cl uk house zaa3? sfp Int.: ‘W ho is person such that s/he cannot B UY this house onl y ?’ b. * H ai co p maai buy Aaming Aaming maai - m- hei buy - not -up bin-gaan w hich-cl uk house zaa3? sfp Int: ‘W hich is the house such that Aaming cannot B UY it onl y ?’ c. * ( cf. S. P . Cheng 2015 , p.169) Aaming Aaming maai - m- hei buy - not -up bin-gaan w hich-cl uk house zaa3? sfp Int.: ‘W hich is the onl y house such that Aaming cannot buy it?’ F inall y , dislocation cop ying of v erbs do not exhibit interv ention eff ects due to the pr esence of wh - expr essions. Both sentences in ( 115 ) ar e acceptable no matter w hether the wh -expr ession is in the the pr e- v erbal or in the post - v erbal position. 43. The incompatibility betw een exhaustiv e/ r estrictiv e f ocus and wh -expr essions has been noted in M andarin (S.-S. Kim 2002a , 2002b , 2006 ; Soh 2005 ; Y ang 2008 , 2012 ; Li and Cheung 2012 , 2015 ). Indeed, S.-S. Kim ( 2006 ) and Y ang ( 2012 ) further note that additiv e f ocus is also incompatible with wh -expr essions in M andarin, but Cantonese is diff er ent in this r egar d, giv en the acceptability of ( 113 ) and the f ollo wing se ntence in (i). (i) A dditiv e f ocus and wh -expr essions lin ev en Aaming Aaming dou also soeng w ant maai buy matje? w hat ‘W hat is the thing such that ev en Aaming also w ants to buy it?’ 86 3.5. FOCUS INTER VENTI O N EFFECTS (115) S ubject v s. object wh -expr essions in dislocation cop ying of v erbs a. B ingo w ho gammaan tonight hoji can fan sleep baat -go eight -cl zung hour aa3 sfp hoji ? can ‘W ho can sl eep f or eight hours tonight?’ b. Aaming Aaming gammaan tonight hoji can fan sleep geinoi ho w .long aa3 sfp hoji ? can ‘H o w l ong can Aaming sleep tonight?’ T a ble 3.8 sums up the discussion abo v e: f ocused elements in the subject/ pr ev erbal position induce interv ention eff ects in thr ee out of f our v erb doubling constructions. In (i) topic, (ii) ‘ ev en ’ - f ocus, and (iii) copula f ocus constructions of v erbs, the r elation betw een the v erb in the perip heral position and the v erb in the base position is disrupted b y f ocused elements in the pr ev erbal position. H o w ev er , no similar interv ention eff ects ar e observ ed in dislocation cop ying of v erbs. T ype Pr ev erbal dak - f ocus Pr ev erbal wh -expr essions T opic - V *( 106 ) *( 112 ) E v en- V *( 108 ) *( 113 ) Copula- V *( 110 ) N/ A ( cf. ( 114 )) DC- V O K ( 111 ) O K ( 115 ) T able 3.8: The interv ention eff ects observ ed with v erb doubling constructions Indeed, f ocus interv eners ar e not r estricted to dak -associates and wh -expr essions. S ubjects asso- ciated with lin ‘ ev en ’ or the copula hai ar e also potential interv eners, as sho wn in ( 116 ). (116) Interv ention eff ects trigger ed b y other f ocused elements a. * Soeng , w ant lin ev en Aaaming Aaming dou also hai co p m- soeng not - w ant heoi go M eigw ok. US Int.: ‘E v en Aaming didn ’t w ant to go to the US. ’ 87 3.5. FOCUS INTER VENTI O N EFFECTS b. * Lin ev en mong look hai co p Aaming Aaming dou also m-gam not -dar e mong look Aafan. Aafan Int.: ‘Onl y Aaming does not ev en dar e to LOO K at Aafan. ’ c. * H ai co p dim touch lin ev en Aaming Aaming dou also m-gam not -dar e dim touch ni - zek this-cl dungmat. animal ‘E v en Aaming does not dar e to T O U CH this animal onl y . ’ F urthermor e, the interv eners need not be in the subject position to block v erb doubling. Elements associated with dak in the pr ev erbal position also lead to interv ention eff ects. In all the sentences in ( 117 ), the v erb doubling constructions ar e block ed b y a non-subject element associated with the pr ev erbal dak . N ote that the sentences ar e w ell - f ormed in the absence of these elements. (117) Interv ention eff ects trigger ed b y f ocused elements in non-subject position a. * Soeng w ant dak onl y ni -bun this-cl syu book Aaming Aaming hai co p soeng w ant maai. buy Int.: ‘ Aaming W ANTS to buy onl y this book. ’ b. * Lin ev en mong , look dak onl y ni -go this-cl jan, person Aaming Aaming dou also m-gam not -dar e mong . look Int.: ‘Onl y this person is such that Aaming does not ev en dar e to LOO K at him/her . ’ c. * H ai co p gong , speak dak onl y J atman, J apanese Aaming Aaming gong - m-dou speak - not -able ze1. sfp Int.: ‘Onl y J apanese is such that Aaming cannot SP EAK onl y . ’ This allo w s us to generalize the interv ention eff ects observ ed in v erb doubling constructions in ( 118 ). 88 3.5. FOCUS INTER VENTI O N EFFECTS (118) Interv ention eff ects in v erb doubling constructions, part 2 ( to be expanded) F ocused elements cannot interv ene betw een the v erb in the perip heral position and the v erb in the base position in (i) topic, (ii) ‘ ev en ’ - f ocus, and (iii) copula f ocus constructions of v erbs. V perip hery ... 8 > < > : Pr e v erbal positi on *F ocused elements 9 > = > ; ... V base ... 8 > < > : P ostv erbal positi on O K F ocused elements 9 > = > ; 3.5.3 N o interv enti on b y quantifi cati onal el ements T o delimit the range of interv ening elements in v erb doubling constructions, I discuss quantificational elements in v erb doubling constructions. They ar e interv eners in wh -questions in v ol ving wh -adv erbs in M andarin (Soh 2005 ; Y ang 2008 , 2012 ), and in other p henomena (Rizzi 2001 , 2004 ). I sho w ele- ments such as existential mark ers, nominal quantifiers, modal v erbs and negation, but no interv ention eff ects ar e observ ed in an y of the v erb doubling constructions. The sentences in ( 119 ) in v ol v e existential sentences in v erb doubling constructions. the existential v erb jau ‘ha v e’ and its associating nominals can occur in the pr ev erbal position without triggering an y interv ention eff ects. (119) Existential jau -sentences in v erb doubling constructions a. Soeng w ant jau ha v e saam-go thr ee-cl jan person hai co p soeng w ant heoi go M eigw ok US ge2. sfp ‘Ther e ar e thr ee peop le w ho w ant to go to the US. ’ b. Lin ev en haang w alk jau ha v e hoksaang student dou also haang - m-dou w alk - not -able sap ten fanzung. minute ‘Ther e ar e students w ho cannot ev en W ALK f or ten minutes. ’ 89 3.5. FOCUS INTER VENTI O N EFFECTS c. H ai co p maai buy jau ha v e di cl jan person maai - m- hei buy - not -up ni -gaan this-cl uk hose ze1. sfp ‘Some peop le cannot aff or d B UYIN G this house onl y . ’ d. J au ha v e jan person gammaan tonight hoji can fan sleep ni - zoeng this-cl cong bed aa3 sfp hoji . can ‘Someone can sleep on this bed tonight. ’ As f or other quantificational elements, I include both subject quantifiers ( of v arious kinds ) and modal v erbs in the sentences in ( 120 ). Again, no interv ention eff ects ar e observ ed. N ote that ( 120 a/ c ) also in v ol v e pr ev erbal negation (i.e., the pr efixal m - and the negativ e v erb mou ‘not.ha v e’), but they do not cause interv ention eff ects either . (120 ) S ubject quantifiers, modal v erbs and negation in v erb doubling constructions a. Soeng w ant cyunbou ev ery hoksaang student dou all jinggoi should hai co p m- soeng not - w ant heoi go M eigw ok US ge2. sfp ‘E v ery students p r obabl y don ’t w ant to go to the US. ’ b. Lin ev en haang w alk houdou man y jan person dou also honang possible haang - m-dou w alk - not -able sap ten fanzung. minute ‘M an y peop le cann ot ev en W ALK f or ten minutes. ’ c. H ai co p maai buy mou not.ha v e jan person gam dar e maai buy ni -gaan this-cl uk house ze1. sfp ‘N o one dar e B UY this house onl y . ’ d. Daaiboufan majority jan person gammaan tonight dou all jatding necessary hoji can fan sleep baat -go eight -cl zung hour aa3 sfp hoji . can ‘The majority of the peop le must be able to sleep f or eight hours tonight. ’ S umming up, v arious quantificational elements can fr eel y occup y t he subject/ pr ev eral position 90 3.5. FOCUS INTER VENTI O N EFFECTS betw een the tw o v erbs in v erb doubling constructions, in contrast to f ocused elements. (121) (The absence of ) interv ention eff ects in v erb doubling constructions, part 3 ( to be expanded) V perip hery ... 8 > < > : Pr e v erbal positi on O K quantificaitonal elements 9 > = > ; ... V base ... ( Object ) 3.5.4 Interim summary T aking stock, it w as first sho wn that v erb doubling constructions tolerate interv ening heads betw een the tw o v erbs. Then it w as r ev ealed that v erb doubling constructions disallo w interv ening f ocus ele- ments in the subject or pr ev erbal positions. On the other hand, quantificational elements in the same position do not induce similar interv ention eff ects. A summary is giv en in ( 122 ). (122) F ocus Interv ention Eff ects in v erb doubling constructions (final) F ocused elements cannot interv ene betw een the v erb in the perip heral position and the v erb in the base position in (i) topic, (ii) ‘ ev en ’ - f ocus, and (iii) copula f ocus constructions of v erbs. V perip hery ... 8 > > > > > > > < > > > > > > > : Pr e v erbal positi on O K heads *f ocused elements O K quantificaitonal elements 9 > > > > > > > = > > > > > > > ; ... V base ... ( Object ) Put diff er entl y , the interv ention eff ects observ ed with v erb doubling constructions should be characterized as f ocus intervention eff ects ( cf. S.-S. Kim 2002a , 2002b , 2006 ; Soh 2005 ; Y ang 2008 , 2012 ; Li and Cheung 2012 , 2015 ), but not quantifier intervention eff ects ( cf. Linebarger 1987 ; Beck 1996 , 2006 ; Beck and Kim 1997 ). 91 3.6. P R O POSAL: HEAD M O VEMENT T O THE SP ECIFIER POSITI O N 3.6 Pr oposal: head mo v ement to the specifi er positi on In view of the empirical observ ations pr esented in pr evious sections, I pr opose that in all f our v erb doubling constructions, v erb mo v ement (head mo v ement ) is in v ol v ed in their deriv ation. All of them ar e trigger ed b y discourse f eatur es, and the v erb mo v es to a specifier position of a functional head in the CP perip hery . I assume the minimalist framew ork of Chomsky ( 2000 , 2001 ), w her e f eatur e checking is achiev ed via an Agr ee operation. I also adopt the suggestion in P esetsky and T orr ego ( 2007 ) that f eatur e in- terpr etability and f eatur e v aluation ar e independent of each other , i.e., uninterpr etable f eatur es ma y enter the syntax being v alued or un v alued, and vice v ersa. 44 (123) F eatur e checking/ v aluation and the operation Agr ee under a Pr obe-Goal sy stem a. An unintr epr etable f eatur e F on a pr obe sear ches its c -command domain and Agr ees with another instance of F , a goal . b. Agr ee betw een a pr obe and a goal is based on a M atching or non-distinctness r elation (i.e., f eatur e identity independentl y of v alue ). c. Agr ee is subject to the locality condition of closest c -command. d. The v alue of a pr obe or a goal is assigned to the other . e. The uninterpr etable f eatur e on the pr obe is deleted upon Agr ee f or LF con v ergence . F urthermor e, I adopt a sp lit CP framew ork after Rizzi ( 1997 ), w her e the CP pr ojection contains a number of functional pr ojections r esponsible f or discourse inf ormation and clause-typing. The pr e- cise components and or derings of these pr ojections will be detailed in the pr oposal. 44. In the pr oposal in Chomsky ( 2001 , p.5), it is suggested that “the uninterpr etable f eatur es, and onl y these, enter the deriv ation without v alues, and ar e distinguished fr om interpr etable f eatur es b y virtue of this pr operty . ” P esetsky and T orr ego ( 2007 ) dev elops a theory of f eatur e v aluation and interpr etability that r elax es this v aluation- interpr etability biconditional. 92 3.6. P R O POSAL: HEAD M O VEMENT T O THE SP ECIFIER POSITI O N 3.6.1 Details of the pr oposal Ther e ar e thr ee components in the pr oposal: (i) f eatur e specification, (ii) f eatur e distribution, and (iii) head mo v ement. F irst, I suggest that ther e ar e tw o discourse f eatur es that can be associated with syntactic constituents, namel y the [F ocus ] f eatur e and the [Def ocus ] f eatur e. 45 F urthermor e, I suggest that the uninterpr etable [F ocus ]/[Def ocus ] f eatur es (hencef orth [ u F ocus ]/[ u Def ocus ]) enter the syntax with a specific v alue, w her eas the interpr etable counterparts (i.e., [ i F ocus ]/[ i Def ocus ]) ar e un v alued ( cf. the f eatur e sy stem in P esetsky and T orr ego 2007 ). The v alue f or [ u F ocus ] can be Co n f or contrastiv e, Ad d f or additiv e, and Exh f or exhaustiv e, but ther e is onl y one v alue f or [ u Def ocus ], i.e., Def f or def ocus. The f eatur e specification is summarized in T able 3.9 . F eatur e U ninterpr etable f eatur e Interpr etable f eatur e F ocus f eatur e [ u F ocus: Co n/ Ad d/Exh] [ i F ocus: _] Def ocus f eatur e [ u Def ocus: Def] [ i Def ocus: _] T able 3.9: F eatur e specification of the f ocus and def ocus f eatur es in Cantonese As f or the distribution of these f eatur es, I suggest that the interpr etable f eatur es ar e assigned to constituents ( e.g., v erbs in our cases of v erb doubling constructions ), and that the uninterpr etable f eatur es ar e held b y diff er ent functional heads in the CP domain. 46 Specificall y , I pr opose that ther e ar e diff er ent p honological r ealizations of the F ocus heads, under diff er ent f eatur e specifications. Lastl y , I suggest that mo v ement of a constituent is independentl y trigger ed b y an EP P f eatur e on the f ocus/ def ocus heads. This EP P f eatur e triggers mo v ement upon successful establishment of an Agr ee r elation betw een a pr obe and a goal. The mo v ement of the goal targets the specifier position 45. This amount to sa ying that that def ocused interpr etation should be tr eated on a par with f ocus interpr etation, in the sense that both notions ha v e syntactic manif estations ( as discourse f eatur es ). This suggestion is def ended in Lee ( 2020 ) based on cr oss- linguistic evidence. S imilar f eatur es ha v e been pr oposed f or right dislocation in J apanese (T akano 2014 ) and scrambling in modern Afrikaans and W est Germanic languages (M olnárfi 2002 ). A slightl y diff er ent suggestion is pr oposed in Lai ( 2019 ), w her e a structur e-building f eatur e attracts a constituent to a position that is interpr eted as giv en inf ormation in dislocation cop ying. I do not further distinguish a def ocus f eatur e and a structur e-building f eatur e r espon- sible f or attracting giv en inf ormation. 46. This is similar to the suggestion in Li and Cheung ( 2012 , 2015 ) that f ocus particles bear uninterpr etable f eatur es, w her eas f ocused p hrases bear interpr etable f eatur es. H o w ev er , this is diff er ent fr om the suggestion in S.-S. Kim ( 2006 ), f or examp le. F or discussions, see § 3.6.3 . 93 3.6. P R O POSAL: HEAD M O VEMENT T O THE SP ECIFIER POSITI O N Construction H ead in the CP domain F eatur e Phonological r ealization T opic - V F ocus [ u F ocus: Co n] null ( ø ) ‘E v en ’ - V F ocus [ u F ocus: Ad d] lin or null ( ø ) Copula- V F ocus [ u F ocus: Exh] hai DC- V Def ocus [ u Def ocus: Def] null ( ø ) T able 3.10 : Distribution of the uninterp r etable f ocus/ def ocus f eatur es and their r ealizations of the pr obe. App l ying this to the v erb doubling constructions in Cantonese, the v erbs bearing the interpr etable f eatur es w ould mo v e to the specifier position of the F ocus head or the Def ocus head in the CP domain due to the pr esence of an EP P f eatur e. This amounts to the suggestion of head-to- specifier mo v ement in the sense of T o y oshima ( 2000 , 2001 ), M atushansky ( 2006 ), V icente ( 2007 ), and H arizano v ( 2019 ). 47 T w o r emarks ar e in or der . F irst, it should be noted that the pr oposal does not pr o vide an ex - p lanation of the doubling eff ects in v erb doubling constructions. I assume that the v erb mo v ement chains ar e ex empted fr om the cop y deletion operation ( under the cop y theory of mo v ement, Chomsky ( 1995b , et seq. )). See some r elev ant discussions in § 3.6.3.2 . Chapter 5 is dedicated to this issue. Second, I str ess that the pr oposal her e is not r estricted to v erb doubling constructions. I t also can r eadil y be extended to the p hrasal counterparts of the v erb doubling constructions in ( 27 ) as discussed in § 3.1 , if the r ela v ent f eatur es ar e associated with p hrasal elements. I r eturn to this issue in § 3.8.2 . 47. A technical concern on head-to-specifier mo v ement ( as opposed to the head-to- head adjunction appr oach (Bak er 1988 , et seq. )) is that it violates the Chain U nif ormity Condition ( CU C), as stated in (i). (i) Chain U nif ormity Condition ( CU C) (Chomsky 1995a , p.406) A chain is unif orm with r egar d to p hrase structur e status. If p hrase structur e status is to be construed as the lev el of pr ojection, the issue w ith head-to-specifier mo v ement is that a head, w hich is pr ojecting, mo v es to a specifier position and becomes non-pr ojecting. In view of this, a str ong thesis is to abandon CU C entir el y (N unes 1998 ; T o y oshima 2000 , 2001 ). A w eak thesis, f ollo wing F ukui and T akano ( 1998 ), is to suggest that CU C onl y r equir es non-distinctness , instead of unif ormity , with r egar d to p hrase structural status. S ubstantiall y , the head in its base position and the head in the specifier ar e non-distinct in the sense that both ar e not pr ojected elements, despite the abo v e noted diff er ence. Indeed, the head-to- head adjunction appr oach to head mo v ement also r equir es similar r elaxation of CU C. I suggest that CU C cannot be maintained in its original f orm, although I do not commit m y self to ho w it should be r evised. F or further justification of head-to-specifier mo v ement, see T o y oshima ( 2001 ), S urán yi ( 2005 ), M atushansky ( 2006 ), and V icente ( 2007 ), among others. 94 3.6. P R O POSAL: HEAD M O VEMENT T O THE SP ECIFIER POSITI O N 3.6.2 An illustr ati on of the pr oposal A schematic r epr esentation of the pr oposal is giv en in ( 124 ), w hich r eflects a particular step of the deriv ation in v erb doubling constructions (but not the ultimate structur e ). This is intended to illustrate the time w hen the pr obes (i.e., the F ocus head and the Def ocus head) ha v e successfull y located their M a tching goals in the c -command domain. Specificall y , in v erb doubling constructions, the first step is that the pr obe, [ u F ocus ] or [ u Def ocus ], finds the matching f eatur e [ i F ocus: _] or [ i Def ocus: _] on the V in the VP . Then, the pr obe Agr ees with its goal ( v alue assignment and f eatur e deletion not sho wn ). Due to the pr esence of an EP P f eatur e of the F ocus/Def ocus head, the V mo v es to the specifier of the corr esponding head. (124) A schematic r epr esentation of the pr oposed head-to-specifier mo v ement F or ce P F or ce sfp Def ocus P Spec V [ i Def ocus: _] Def ocus’ Def ocus ø [ u Def ocus: Def , EP P ] F ocus P Spec V [ i F ocus: _] F ocus’ F ocus 8 < : ø [ u F ocus: Co n , EP P ] lin [ u F ocus: Ad d, EP P ] hai [ u F ocus: Exh , EP P ] 9 = ; TP VP { V [ i F ocus: _ ] V [ i Def ocus: _] } T w o qualifications ar e in or der . F irst, the structur e in ( 124 ) is a halfw a y deriv ation of v erb dou - bling constructions. The deriv ation continues with a subsequent p hrasal mo v ement into the specifier position of F or ce P . S ince F or ce P is headed b y sentence- final particles, I r ef er to this mo v ement as SFP - 95 3.6. P R O POSAL: HEAD M O VEMENT T O THE SP ECIFIER POSITI O N driven movement f or con v enience. 48 B y w a y of illustration, in topic constructions, ‘ ev en ’ - f ocus con- structions and copula f ocus constructions, the F ocus P undergoes this SFP -driv en mo v ement to Spec F or ce P , deriving the v erb- initial or der , as depicted in ( 125 a ). On the other hand, in dislocation cop y - ing, the TP (but not the Def ocus P ) undergoes the SFP -driv en mo v ement, stranding the v erb in Spec Def ocus P , as depicted in ( 125 b ). N ote that the pr oposed structur e in ( 124 ) allo w s the co-occurr ence of Def ocus P and F ocus P . In such case, F ocus P undergoes SFP -driv en mo v ement, stranding ( again ) the Def ocus P , as depicted in ( 125 c ). (125) SFP -driv en mo v ement ( to the specifier position of F or ce P ) a. [ F or ce P [ F ocus P V [ i F ocus ] ...] [ F or ce’ sfp t F ocus P ] ] F ocus P mo v ement b. [ F or ce P [ TP ... ] [ F or ce’ sfp [ Def ocus P V [ i Def ocus ] t TP ] ] TP mo v ement c. [ F or ce P [ F ocus P V [ i F ocus ] ...] [ F or ce’ sfp [ Def ocus P V [ i Def ocus ] t F ocus P ] ] ] F ocus P mo v ement T o see examp les of ( 125 c ), consider the sentences in ( 126 ). In ( 126 a ), the v erb maai ‘buy’ is topical - ized (i.e., it bears the [ i F ocus ] f eatur e ), w her eas the modal v erb jinggoi ‘ should’ is right -dislocated and doubled (i.e., it bears the [ i Def ocus ] f eatur e ). The SFP -driv en mo v ement strands the Def ocus P . S imilar can b e said to ( 126 b ), w her e the v erb sik ‘ eat’ is f ocused and the v erb jiu ‘w ant’ is def ocused. (126) Co-occurr ence of Def ocus P and F ocus P a. M aai , buy Aaming Aaming jinggoi should hai co p soeng w ant maai buy ge2 sfp jinggoi . should ‘ As f or buying, Aaming pr obabl y w ants to buy this book. ’ 48. I r eturn to the justification of this mo v ement shortl y . 96 3.6. P R O POSAL: HEAD M O VEMENT T O THE SP ECIFIER POSITI O N b. Lin ev en sik eat Aaming Aaming dou also jiu w ant sik eat zeoi most gai expensiv e ge m o d je thing aa3 sfp jiu w ant ‘ Aaming e v en w ants to EA T the most expensiv e thing. ’ T urning to the natur e of the SFP -driv en mo v ement, it is b y no means an ad hoc mo v ement op- eration to deriv e the w or d or der in v erb doubling constructions. I t is a commonl y assumed step f or pr oposals that adopt a head- initial anal y sis of sentence- final particles in Chinese. The general idea is that in or der to deriv e the final position of sentence- final particles, the main clause must mo v e t o a higher position, f ollo wing the spirit of the Linear Corr espondence Axiom (Ka yne 1994 ). F or examp le, in a r egular declarativ e sentence, SFP - mo v ement is still needed, schematized in ( 127 ). (127) SFP - mo v ement, in the absence of Def ocus P and F ocus P [ F or ce P [ TP S V O ] [ F or ce’ sfp t TP ] ] This mo v ement step is not onl y pr oposed or assumed in w orks on right dislocation in L. Y .-L. Cheung ( 2009 ), Lee ( 2017 ), W ei and Li ( 2018 ), and Lai ( 2019 ), but also in w orks on the sentence- final particles, such as T ang ( 1998b ), S ybesma ( 1999 ), S impson and W u ( 2002 ), Lin ( 2010 ), H sieh and S ybesma ( 2011 ), and P an ( 2020 ) ( see also S impson ( 2014 ) f or o v erview and r ef er ences ther ein ). In other w or ds, the SFP -driv en mo v ement is a step needed independentl y of v erb doubling constructions. Second, f or ‘ ev en ’ - f ocus constructions and copula f ocus constructions, ther e ar e or dering issues with r egar d to the surface position of the f ocus particles lin and hai and their f ocus associates. In the structur e in ( 124 ), the w or d appears to be V - lin/hai , contrary to facts. I suggest that both lin and hai ar e pr efix es in need of p honological support, and that they ar e pr efix ed on their f ocus associate in the specifier position upon Spell -Out such that they appear to the left of their f ocus associates. 49 49. Another possibility is that the f ocus particles, lin and hai , undergo a short head mo v ement to a position abo v e it s specifier . This suggestion is made in the anal y sis of dak ‘ onl y’ in T ang ( 2002 ). This option is p lausible under the assumption that dak is a genuine v erb and the head mo v ement is indeed short v erb mo v ement that is common in Chinese (H uang 1994 ). I do not adopt this possibility , since I suggest that lin and hai ar e diff ernt fr om dak in this r egar d, and I tr eat them as functional categories, f ollo wing Sh yu ( 1995 ) and C. C.-H. Cheung ( 2008 , 2015 ). 97 3.6. P R O POSAL: HEAD M O VEMENT T O THE SP ECIFIER POSITI O N 3.6.3 Deri ving the pr operti es of v erb doub ling constructi ons 3.6.3.1 The or dering of the functi onal pr ojecti ons in the CP perip hery R ecall the section summary in § 3.3.3 , r epeated belo w in T able 3.11 . In the pr oposal, I suggested that a F ocus head is in v ol v ed in thr ee out of f our v erb doubling constructions, with the ex ception that a Def ocus head is designated f or dislocation cop ying. I suggest that that this distinction is made based on w hether the v erb in the r elev ant constructions in v ol v e a contrastiv e interpr etation or not. T ype Discourse eff ects Contrastiv eness R elativ e or der with topics T opic - V contrastiv e topic 4 T opic > V / *V > T opic ‘E v en ’ - V additiv e f ocus 4 T opic > V / *V > T opic Copula- V exhaustiv e f ocus 4 T opic > V / *V > T opic DC- V def ocused/ giv en elements 6 *T opic > V / ??V > T opic T able 3.11 : The discourse eff ects of the v erb doubling constructions ( r epeated) Also, I ha v e assumed that Def ocus P is structurall y higher than the F ocus P . In eff ect, this en- sur es that v erbs that r eceiv e a contrastiv e r eading appear sentence- initiall y ( w hen F ocus P undergoes the SFP -driv en mo v ement ), w her eas those that do not appear sentence- finall y ( w hen the SFP -driv en mo v ement strands Def ocus P ). A r emaining issue is the r elativ e or der betw een Def ocus P /F ocus P and discourse topics. The r ela- tiv e or der betw een the v erbs and topics in T able 3.11 f ollo w s if w e assume a T opic P that is sandwiched betw een Def ocus P and F ocus P . The clausal structur e in the CP domain is depicted in ( 128 ). (128) The pr oposed left perip hery in Cantonese F or ce P > Def ocus P > T opic P > F ocus P > TP This straightf orw ar dl y exp lains w h y a discourse topic must pr ecede a v erb in the left perip hery: because T opic P is structurall y higher the F ocus P . On the other hand, it also accounts f or w h y a topic cannot pr ec ded a v erb in the right perip hery: because Def ocus P is higher than T opic P . As f or w h y a def ocused v erb cannot pr ecede a topic, I suggest that Def ocus P has to be stranded w hen SFP -driv en 98 3.6. P R O POSAL: HEAD M O VEMENT T O THE SP ECIFIER POSITI O N mo v ement app lies to the structur e, w her eas T opic P must undergo SFP -driv en mo v ement. 50 N ote that the suggestion her e is consistent with the topograp h y of CP perip hery put f orth f or Cantonese in C. C.-H. Cheung ( 2015 ), w her e the f ocus “field” is located belo w the topic “field” ( cf. Benincà and P oletto 2004 ). W ith the intr oduction of Def ocus P , it seems p lausible to suggest that both Def ocus P and T opic P fall into the topic “field, ” since they in v ol v e non-contrastiv e interpr etations. 3.6.3.2 The mo v ement pr operti es in v erb doub ling constructi ons The lexical identity eff ects discussed in § 3.4.1 f ollo w fr om a mo v ement anal y sis, particularl y fr om the cop y theory of mo v ement (Chomsky 1995b ; N unes 1995 , 2004 ). U nder this theory , mo v ement chains cr eate tw o identical copies of the mo ving elements (instead of lea ving behind traces ). U su - all y , the lo w er cop y is deleted b y a mechanism (i.e., cop y deletion ) in the interface sy stem, and onl y the higher cop y surviv es. H o w ev er , it has been suggested that the mechanism of cop y deletion is not har d- wir ed and can be disrupted or suspended due to independent r easons (Bošk o vić 2007 ; N unes 2011 ). I suggest that this is the case f or v erb doubling constructions. In other w or ds, instead of tak - ing “ gap less” structur es as evidence against mo v ement, I suggest that “ gap less” structur es with lexical identity eff ects ar e indeed evidence f or mo v ement. The ex ceptional r ealization of the lo w er cop y is due to independent constraints in the interface sy stem r esponsible f or linearization. F or a detailed pr oposal on the doubling eff ects in v erb doubling constructions, see Chapter 5. The summary of the locality eff ects with r egar d to (im )penetrability is r epeated in ( 129 ). W hile certain syntactic configurations constitute syntactic “islands” and ar e inaccessible to mo v ement op- erations, CP boundaries do not constitute syntactic islands. 50. I t should be ackno w ledged that the basis of this suggestion is entir el y empirical. T o the extent that the pr ecise natur e of SFP -driv en mo v ement is not entir el y clear in the literatur e, I do not attempt a deeper exp lanation on this pr operty of Def ocus P . 99 3.6. P R O POSAL: HEAD M O VEMENT T O THE SP ECIFIER POSITI O N (129) A schematic r epr esentation of the locality eff ects in v erb doubling constructions V perip hery ... 8 > < > : CP boundaries *I sland boundaries 9 > = > ; ... V base S uch a locality pr ofile is not unique to v erb doubling constructions in Cantonese, but has been r e- ported in man y other cases of v erb mo v ement (K oopman 1984 ; V icente 2007 ; H arizano v and Grib- ano v a 2019 , among man y others; see discussions in § 3.2 ). T o the extent that such a locality pr ofile is similar to typical instances of A ’ - mo v ement (Chomsky 1973 , et seq. ), I assume that the v erb mo v ement in the long -distance v erb doubling constructions pr o- ceeds in a successiv e cy clic fashion, and stops at an intermediate position, i.e., the specifier position of CP , bef or e exiting a CP . This is illustrated in ( 98 ) ( cf. long -distance A ’ - mo v ement (Chomsky 2000 ; 2001 , among man y others )). (130 ) A schematic r epr esentation of long -distance v erb doubling constructions [ F ocus P /Def ocus P V perip hery [ TP ... [ CP V int e r mediate C [ TP ... V base ... ] ] ] ] 3.6.3.3 A s yntacti c e xp lanati on to F ocus Inter v enti on Eff ects In § 3.5 , I arriv ed at the generalization of F ocus Interv ention Eff ects, r epeated in ( 131 ). (131) F ocus Interv ention Eff ects in v erb doubling constructions (final) F ocused elements cannot interv ene betw een the v erb in the perip heral position and the v erb in the base position in (i) topic, (ii) ‘ ev en ’ - f ocus, and (iii) copula f ocus constructions of v erbs. 100 3.6. P R O POSAL: HEAD M O VEMENT T O THE SP ECIFIER POSITI O N V perip hery ... 8 > > > > > > > < > > > > > > > : Pr e v erbal positi on O K heads *f ocused elements O K quantificaitonal elements 9 > > > > > > > = > > > > > > > ; ... V base ... ( Object ) I suggest that F ocus Interv ention Eff ects ar e a natural consequence of the locality condition of Agr ee (Chomsky 2000 , 2001 ), w her e Agr ee is subject to the locality condition of closest c -command. I il - lustrate this idea with the schematic structur e in ( 132 ). I omitted the F or ce P and the EP P f eatur es f or their irr elev ance. (132) A configuration of F ocus Interv ention Eff ects in v erb doubling constructions Def ocus P Def ocus ø [ u Def ocus: Def ] F ocus P F ocus 8 < : ø [ u F ocus: Co n ] lin [ u F ocus: Ad d ] hai [ u F ocus: Exh ] 9 = ; ... XP [ i F ocus: _] VP { V [ i F ocus: _ ] V [ i Def ocus: _] } 2 nd matching Goal 1 st matching Goal In ( 132 ), an f ocused element, XP , interv enes betw een the F ocus head and the v erb. Both the XP and the v erb bear the [ i F ocus: _] f eatur e. A t the time w hen the F ocus head sear ches its c -command domain f or a matching f eatur e, it encounters the structurall y closer XP , bef or e the v erb in the VP . U nder the locality conditions of Agr ee, the F ocus head must Agr ee with XP instead of the v erb. In other w or ds, the v erb cannot be successfull y targeted f or mo v ement in the pr esence of a structurall y higher f ocused element. In contrast, no interv ention eff ects ar e observ ed if the f ocused element is in the object/ postv erbal 101 3.6. P R O POSAL: HEAD M O VEMENT T O THE SP ECIFIER POSITI O N position, since it does not interv ene betw een the F ocus head and the v erb. Also, a v erb or a quan- tificational element w ould not interv ene if they do not possess the [ i F ocus:_]. A dditionall y , F ocus Interv ention Eff ects do not arise in dislocation cop ying, since the [ i F ocus:_] on the XP is not a match- ing f or the Def ocus head. The [ i Def ocus:_] on the v erb is the first matching Goal f or the Def ocus head. This deliv ers the F ocus Interv ention Eff ects. This exp lanation amounts to a f eatural characterization of the interv ener f or head mo v ement. M or e generall y , this can be r egar ded as a natural extension of the f eatur e-based r elativized minimality discussed in Rizzi ( 2001 , 2004 ), w hich app lies to both head and p hrasal mo v ement. 51 The deriv ation in ( 132 ) w ould be w ell - f ormed if the F ocus head Agr ees with XP , instead of the v erb. F or examp le, in ( 133 a ), the f ocus head lin can be associated with a pr ev erbal wh -expr ession. 52 In ( 133 b ), the f ocus head hai is associated with dak and its associate. 53 54 (133) F ocus heads Agr ee with interv ening f ocused elements a. Lin ev en bingo w ho dou also tai -gw o r ead-exp ni -bun this-cl syu? book ‘E v en WH O has r ead this book?’ b. H ai co p dak onl y Aaming Aaming m-gam not -dar e dim touch ni - zek this dongmat animal ze1. sfp ‘Onl y AAMIN G does not dar e to touch this animal. ’ I t should be r emark ed that the suggestion her e amounts to a syntactic account to interv ention eff ects along the line of Rizzi ( 2001 , 2004 ), S.-S. Kim ( 2006 ), Y ang ( 2008 , 2012 ), and Li and Cheung 51. A similar f eatur e-based appr oach to constrain head mo v ement is suggested in R oberts ( 2001 , p.140-145), w her e he discusses Long H ead M o v ement in B r eton. This possibility is also mentioned in passing in T o y oshima ( 2001 ). 52. S imilar cases in v ol ving double association with f ocus and question operators ar e discussed in Li and Cheung ( 2012 , 2015 ). 53. ( 133 b ) raises an issue of ho w double exhaustiv e f ocus marking is interpr eted compositionall y , on w hich I am ag - nostic. H o w ev er , as far as syntactic dependencies ar e concerned, this configuration w ould be allo w ed if dak also bears a [ u F ocus ] f eatur e, and thus both dak and hai Agr ees with the [ i F ocus ] f eatur e on Aaming . 54. In princip le, mo v ement to Spec F ocus P is in v ol v ed, but, as suggested, lin and hai ar e pr efixal elements and w ould be linearized to the left of the f ocused elements, r endering the mo v ement string - v acuous. 102 3.7. AL TERN A TIVE AN AL Y SES T O A HEAD M O VEMENT AP P R O A CH ( 2012 , 2015 ), instead of a semantic account to interv ention eff ects (Beck 1996 ; 2006 ; Beck and Kim 1997 , among man y others ). The motiv ation f or the f ormer comes fr om the observ ation that f ocus operators b y themsel v es do not cause interv ention eff ects, as discussed in § 3.5.2 . F or examp le, in ( 107 ), r epeated in ( 134 ), zinghai ‘ onl y’ in the pr ev erbal position does not cause interv ention eff ects. 55 (134) F ocus operators do not trigger interv ention eff ects (= ( 107 )) H eoi go Aaming Aaming hai co p jinghai onl y heoi -dak go-onl y jat -go one-cl deif ong. p lace ‘ As f or going, Aaming has been to onl y one p lace. ’ A semantic appr oach to interv ention eff ects w ould pr edict that the lo w er f ocus operator (i.e., zinghai ) w o uld block a higher f ocus operator (i.e., the ( null) F ocus head), since the f ormer consumes all the alternativ es within its local domain, bef or e the latter can locate its f ocus associate. H o w ev er , the sentence is w ell - f ormed. The acceptability of ( 134 ) f ollo w s fr om the curr ent pr oposal, if zinghai does not bear an y [ u F ocus ] f eatur e, and thus does not interv ene betw een the tw o v erbs. 56 3.7 Alternati v e anal y ses to a head mo v ement appr oach Ther e ar e tw o types of alternativ e anal y ses of the kinds of patterns consider ed her e. N on- mo v ement appr oaches might suggest that the tw o v erbs in v erb doubling constructions ar e not ( dir ectl y ) r e- lated b y mo v ement dependencies. Phrasal mo v ement appr oaches w ould suggest that the mo ving con- stituent is not a head but a p hrase. I discuss these tw o appr oaches ( and their v ariants ) in the f ollo wing subsections, r espectiv el y . 55. S imilar examp les include ( 108 ) and ( 110 ), as w ell as dak ‘ onl y’ in ( 133 b ) 56. This amounts to the suggestion that not all f ocus operators bear [ u F ocus ]. One possible exp lanation f or this sp lit is that f ocus operators with [ u F ocus ] trigger mo v ement of the f ocused elements, w her eas those without it do not. This is supported b y that fact that zinghai can occur at a distance with its f ocus associate, i.e., jat-go deif ong ‘ one p lace’ in ( 134 ). I ts contrasts with lin , hai and dak . 103 3.7. AL TERN A TIVE AN AL Y SES T O A HEAD M O VEMENT AP P R O A CH 3.7 .1 N on - mo v ement appr oaches S ince v erb doubling constructions in v ol v e “ gap less” structur es, this pr o vides some initial motiv ation to pursue a non- mo v ement/base generation account. W hile, to the best of m y kno w ledge, no base generation account has been pr oposed f or v erb doubling constructions, their p hrasal counterparts ha v e been independentl y argued to in v ol v e base generation, especiall y in “ gap less cases” . H o w ev er , I will first sho w that a dir ect app lication of a base generation appr oach to v erb doubling constructions fails to captur e the pr operties observ ed in pr evious sections. F or the sak e of argument, I exp lor e a mor e specific v ersion of base generation, w hich is coup led with operator mo v ement. I again sho w that this h ybrid appr oach does not captur e the r elev ant facts and leads to undesirable pr edictions. 3.7 .1.1 Base gener ati on As far as discourse eff ects ar e concerned, each of the v erb doubling constructions has a p hrasal coun- terpart that does not in v ol v e a gap, illustrated in ( 135 ) ( cf. their gapped counterparts in ( 27 )). (135) Examp les of “ gap less” cases a. T opic constructions with an “ aboutness” r elation (Chao 1968 , among others ) Seoigw o fruit ngo I zeoi most zungji lik e monggw o. mango ‘F or f ruits, I lik e mango the most. ’ b. ‘E v en ’ - f ocus constructions with r esumptiv e pr onouns ( cf. Sh yu 1995 , p.139) Lin ev en A aming i Aaming Aafan Aafan dou also m-tai not - r ead k eoi i his ge m o d syu. book ‘ Aafan d idn ’t ev en r ead AAMIN G’ s book. ’ 104 3.7. AL TERN A TIVE AN AL Y SES T O A HEAD M O VEMENT AP P R O A CH c. Copula f ocus constructions with r esumptiv e pr onouns H ai co p ni -go this-cl doujin i dir ector Mingzai Mingzai zeoi most zungji lik e tai w atch k e oi i s/he pak shoot ge m o d dinjing. film ‘Mingzai lik e to w atch the films dir ected b y THIS D IRECT O R the most. ’ ( cf. C. C. -H. Cheung 2015 , p.95, modified) d. Dislocation cop ying with non- identical copies (L. Y .-L. Cheung 2015 , p.230 ) Gam then k eoi i he zau - m- zau lea v e- not - lea v e hou good ne1 sfp F atgw o k F rance l ou i ? guy ‘Then should he, the F r ench guy , lea v e?’ I t is commonl y suggested that these sentences in v ol v e base generation of a nominal p hrase in the pe- rip hery position. The nominal p hrase establishes a non- mo v ement dependency with its associates, w hich ma y be a semanticall y r elated category in ( 135 a ) or a r esumptiv e pr onoun in ( 135 b-d). 57 Ap- p l ying this idea to v erb doubling constructions, one ma y posit that a v erb is base generated in the perip heral position and is associated with the v erb in the base position via a non- mo v ement depen- dency . As such, no (head) mo v ement is needed. H o w ev er , such a base generation appr oach fails to account f or the pr operties of v erb doubling constructions. F irst, in the cases in ( 135 ), the base generated element need not be lexicall y identical to its associate, as long as they ar e semanticall y r elated. This diff ers substantiall y fr om v erb doubling constructions w hich exhibit the lexical identity eff ects, as discussed in § 3.4.1 . Second, and mor e importantl y , a signatur e pr operty of base generation structur es is their island insensitivity . F or examp le, in the sentences in ( 135 b ) and ( 135 c ), the r esumptiv e pr onoun in both cases is embedded in a comp lex nominal structur e. H o w ev er , no island eff ects ar e observ ed. This is in sharp contrast with v erb doubling constructions, since, they ha v e been sho wn to be sensitiv e to syntactic islands in § 3.4.2 . 58 57. I t is possible that r esumption also in v ol v es mo v ement (f or r ecent discussions of Cantonese, see Y ip and Ahenk orah 2022 ). 58. P an ( 2019 ) argues that copula f ocus constructions ( or ex -situ f ocus cleft constructions, in his terminology ) in M an- darin in v ol v e base generation structur es. His pr oposal is based on the observ ation that ther e is no island sensitivity in 105 3.7. AL TERN A TIVE AN AL Y SES T O A HEAD M O VEMENT AP P R O A CH Lastl y , a base generation does not pr edict F ocus Interv ention Eff ects, as detailed in § 3.5 . I t is unclear ho w the base generation of v erbs in the perip heral position w ould be interv ened b y f ocused elements, giv en the lack of syntactic dependencies betw een the tw o v erbs. 3.7 .1.2 Base gener ati on p lus oper ator mo v ement Bef or e a base generation appr oach is entir el y dismissed, it is instructiv e to note that the issue of is- land sensitivity might be cir cum v ented, if the base generation appr oach is equipped with operator mo v ement, w hose a v ailability in Chinese is independentl y motiv ated. In such an appr oach, it might be suggested that, w hile the v erb in the perip heral position is base generated, mo v ement of a null operator occurs in the lo w er clause. The dependency betw een the tw o v erbs is thus indir ect: it is mediated b y a null operator that establishes a pr edication r elation. Importantl y , island sensitivity f ollo w s fr om the h ypothetical operator mo v ement, instead of head mo v ement. Dra wing largel y on the anal y sis pr oposed f or dak ‘ onl y’ in T ang ( 2002 ), I suggest that a conceiv able imp lementation of his idea to v erb doubling constructions consists of tw o steps: first, a null operator ( of the same semantic type as v erbs ) mo v es to the edge of the clause; second, the clause is pr edicated on the base generated v erb in the perip hery . 59 This idea is illustrated in ( 136 ). 60 (136) An illustration of a base generation p lus operator mo v ement anal y sis [ F ocus P V [ F ocus’ F ocus [ TP O P [ TP S ubject ... t O P ... ( Object ) ] ] ] ] wher e O P is o f the same semantic type as verbs 1. O P mo v ement 2. Pr edication the copula f ocus constructions in M andarin. Giv en the crucial diff er ence in empirical observ ations, I suggest that the deriv ation f or copula f ocus constructions ar e diff er ent in Cantonese and M andarin. 59. Operator mo v ements ha v e been independentl y motiv ated on empirical gr ounds, including passiv e constructions ( T ing:1995 ) and r elativ e structur es in v ol ving wh -adjuncts ( e.g., A oun and Li 2003 ). Although these pr oposals ar e pr oposed f or M andarin, as far as I can see, the argumentation largel y be app lied to Cantonese as w ell. 60. This appr oach is indeed similar to a suggestion in def ense of a base generation anal y sis in Cable ( 2004 ) and V icente ( 2007 ), w her e a v erb ma y be base-generated in the edge of the lo w er clause (i.e., Spec CP ). W hen this CP is is embedded in an syntactic island, mo v ement of this base-generated v erb to the matrix perip hery position w ould violate island conditions. In this appr oach, v erb mo v ement is still needed, and the onl y diff er ence with the pr oposed head mo v ement appr oach lies in the base generation position of the v erb. F urthermor e, unlik e the case in Y iddish, v erb doubling constructions in Cantonese disp la y lexical identity eff ects as sho wn in § 3.4.1 , and thus ther e is little motiv ation to posit a base generated v erb at Spec CP . 106 3.7. AL TERN A TIVE AN AL Y SES T O A HEAD M O VEMENT AP P R O A CH As a note on the legitimacy of operator mo v ement, this can be construed as a syntactic corr elate of a semantic operation, namel y , lambda conversion , w hich con v erts a pr oposition/TP into a pr edi - cate that denotes a r elation. Abstracting a w a y fr om irr elev ant details, the semantic denotation of the F ocus P in ( 136 ) can be giv en in ( 137 ). (137) The denotation of the TP in ( 136 ) R [ R(x, y ) ] wher e R denotes a r elation, x is the subject and y the object ( if any ) S ince this h ybrid appr oach in v ol v es mo v ement, it handles the locality eff ects better than a pur e base generation appr oach. I sland sensitivity f ollo w s, since operator mo v ement should be constrained in a similar w a y . I t is expected to violate island conditions if the operator is embedded in the r elev ant structur es and undergoes mo v ement into a higher clause. In or der to deriv e F ocus Interv ention Ef - f ects, it must be assumed that the operator mo v ement under discussion b y natur e is a kind of f ocus mo v ement, such that it is sensitiv e to interv ening f ocused elements. Although it r emains unclear w h y this must be the case (giv en that not all cases of operator mo v ement ar e f ocus- r elated), the assumption does not seem imp lausible. H o w ev er , ther e ar e a number of challenges to this appr oach. F irst, additional assumptions ar e needed to accommodate the lexical identity eff ects. F or examp le, it must be assumed that, at the point of pr edication, t he operator does not onl y absorb the semantic content of the v erb, but also its p hono- logical content ( w hich is then transmitted to its trace/lo w er cop y ), r esulting in v erb doubling. This r epr esents, as far as I kno w , an unpr ecedented w a y to deriv e doubling eff ects, and its v alidity hinges on a general theory of syntactic operators, w hich I cannot do justice her e. A mor e substantial challenge comes fr om the idiom chunks as discussed in § 3.4.3 . U nder this base generation appr oach, the v erb in an idiomatic expr ession enters into the structur e independentl y of the r est of the expr ession. I t is thus pr edicted that the idiomatic r eading should be una v ailable; ho w ev er , this is not the case. 107 3.7. AL TERN A TIVE AN AL Y SES T O A HEAD M O VEMENT AP P R O A CH 3.7 .2 Phr asal mo v ement appr oaches Another alternativ e to the pr oposed head mo v ement is to maintain that ther e ar e mo v ement depen- dencies betw een the tw o v erbs in v erb doubling constructions, but the mo ving element is a p hrase, instead of a head, and w hat occupies the perip heral position is a VP instead of a single V head. As such, w hat appears to be head mo v ement is p hrasal mo v ement in disguise. 61 (138) A h ypothetical p hrasal mo v ement in v erb doubling constructions [ F ocus P [ VP ... V perip hery ... ] F ocus [ TP ... [ VP V base ... ] ] ] VP mo v ement Ther e ar e at least tw o conceiv able w a y s to deriv e a VP that contains onl y a V head, w hich I exp licate ( a nd a rgue against ) in the f ollo wing tw o subsections. 3.7 .2.1 R emnant VP mo v ement One w a y to cr eate a VP that contains onl y a head is to posit independent mo v ement operations that ev acuate the VP bef or e mo v ement of the VP . T o illustrate the idea, imagine a VP that consists of a v erb and its object. Bef or e the VP mo v es to the perip hery position, the object mo v es out of VP f or independent r easons. I t could be an instance of object scrambling or object s hift. S ubsequentl y , w hen the VP mo v es at a later step in the deriv ation, it mo v es as if the v erb alone is mo ving. (139) A h ypothetical r emnant p hrasal mo v ement in v erb doubling constructions, v ersion 1 [ F ocus P [ VP ... V perip hery ... ] F ocus [ TP ... S ubject ... Object 1 [ VP V base t 1 ] ] ] 2. R emnant VP mo v ement 1. Object mo v ement 61. VP mo v ement in the r elev ant constructions is allo w ed, although ther e is no doubling eff ects on the v erbs; see dis- cussions in § 3.8.2 . 108 3.7. AL TERN A TIVE AN AL Y SES T O A HEAD M O VEMENT AP P R O A CH This appr oach is argued to be r esponsible f or pr edicate fr onting in German and R ussian ( see, f or ex - amp le, den Besten and W ebelhuth 1990 ; Abels 2001 , among man y others ), both of w hich has a con- siderabl y pr oductiv e mechanism of ( object ) scrambling. In or der to app l y this appr oach to v erb doubling constructions in Cantonese, tw o issues ha v e to be addr essed. F irst, object mo v ement ( or an y mo v ement operation that ev acuates the VP ) must be independentl y motiv ated f or each case of v erb doubling constructions. 62 This is not imp lausible, since the object ma y be fr onted f or contrastiv e f ocus in Cantonese, giving a SO V w or d or der . 63 If, in each case of v erb doubling constructions, the object (if an y ) is fr onted, then the object in v erb doubling constructions must al w a y s be in contrastiv e f ocus. This is, ho w ev er , not the case. Second, and mor e criticall y , the w or d or der depicted in ( 139 ), i.e., V SO V , is not the surface w or d or der f or v erb doubling constructions. An extra mo v ement step of the VP must be posi ted. P articular , after object mo v ement, the VP must then mo v e to a position bef or e the object. (140 ) A h ypothetical r emnant p hrasal mo v ement in v erb doubling constructions, v ersion 2 [ F ocus P [ VP ... V perip hery ... ] F ocus [ TP ... S ubject ... [ VP V base ... ] Object 1 [ VP V base t 1 ] ] ] 2. 1 st R emnant VP mo v ement 1. Object mo v ement 3. 2 nd R emnant VP mo v ement F urthermor e, it has to be assumed that the lo w est VP has to be deleted w her eas the intermediate VP has to be pr onounced, such that the sentences ar e onl y pr onounced with tw o v erbs but not thr ee in v erb doubling constructions. T o the extent that the mo v ement operation and th Spell -Out mechanism is not independentl y motiv ated, they appear to be ad hoc operations specificall y designed f or v erb doubling construction to deriv e the w or d or der . 62. Hinterhölzl ( 2002 ) suggests that the VP -ev acuating mo v ement can be r egar ded as a licensing movement f or r emnant mo v ement. In other w or ds, these mo v ements need not corr espond to other attested mo v ement operations in the language, and ar e parasitic on r emnant mo v ement. This suggestion is criticized in Landau ( 2006 ) and Cheng and V icente ( 2013 ). I do not pursue this possibility further . 63. But other kinds of object scrambling is highl y r estricted in Chinese, see Soh ( 1998 ) and Cheng and V icente ( 2013 ). 109 3.7. AL TERN A TIVE AN AL Y SES T O A HEAD M O VEMENT AP P R O A CH 3.7 .2.2 VP mo v ement with subsequent del eti on Another w a y to cr eate a VP that contains a single v erb is that the VP mo v es to the periperhal position and then a subsequent deletion operation erases ev erything but the v erb, as depicted in ( 141 ). (141) A h ypothetical p hrasal mo v ement with deletion in v erb doubling constructions [ F ocus P [ VP V perip hery Object ] F ocus [ TP S ubject [ VP V base Object ] ] ] VP mo v ement + object deletion The v alidity of this appr oach, thus, hinges on the v alidity of the pr ecise natur e of this deletion operation. Ther e ar e a f ew possibilities f or this deletion. F irst, since I ha v e assumed the cop y theory of mo v ement and a mechanism of cop y deletion, it might be that the object in ( 141 ) is deleted at the interface component. H o w ev er , this r equir es non-standar d positional deletion, i.e., deletion of the higher cop y , inste ad of the lo w er cop y . 64 F urthermor e, the deletion must be partial , as it deletes onl y a subpart of the w hole VP . 65 Based on the standar d understanding of mechanisms of cop y deletion, I r eject this possibility . Second, the deletion of the object in ( 141 ) might be r egar ded as an instance of argument ellipsis , w her e a missing argument does not lea v e behind a pr onominal, v ariable or nominal trace. 66 H o w ev er , it has been sho wn that the distribution of missing objects is r estricted. F or examp le, ther e ar e v erbs that cannot tak e a null clausal object. I contrast the v erb gu ‘ guess’ and soengseon ‘believ e’ in their ability to license a null clausal object. The contrast first observ ed and discussed in M andarin in Y .-H. A. Li ( 2005 ) and A oun and Li ( 2008 ). 64. This option is not impossible, as discussed as N unes ( 2004 ) and Bošk o vić ( 2007 ), but r equir es independent justifica- tion. 65. P artial deletion is argued to be possible, as in F anselo w and Ća v ar ( 2002 ). Again, it r equir es independent justification on w h y it is selectiv el y emp lo y ed in v erb doubling constructions but not other p henomena. 66. Argument ellipsis is argued to be pr esent in J apanese and K or ean (O ku 1998 ; S. Kim 1999 , among others ). I ts a v ail - ability in Chinese is questionable, as discussed in Y .-H. A. Li ( 2005 ), A oun and Li ( 2008 ), and Y .-H. A. Li ( 2014 ). S ince I argue against this possibility , the pr ecise characterization of this deletion operation does not bear on the discussion. 110 3.7. AL TERN A TIVE AN AL Y SES T O A HEAD M O VEMENT AP P R O A CH (142) Some v erbs disallo w null clausal objects N go I gu/ guess soengseon believ e Aaming Aaming lai. come Aafan Aafan dou also f *gu/ guess soengseon g . believ e ‘I guess/ believ e that Aaming ( will) come. Aafan also guess/ believ e ( so ). ’ If v erb doubling constructions in v ol v e argument ellipsis as a w a y depicted in ( 141 ), it is pr edicted that v erbs lik e gu ‘ guess’ cannot be u sed in v erb doubling constructions, since it does not license a null clausal object. H o w ev er , this is not the case. (143) V erbs that disallo w null clausal objects ar e compatible in v erb doubling constructions Gu guess ngo I zicin bef or e hai co p gu guess Aaming Aaming wui will lai come ge2. sfp Batgw o but k eoi he zeoihau at.last mou not.ha v e lai. come ‘I ha v e guessed that Aaming will come. But he didn ’t come at last. ’ As such, in or der to maintain the pr oposal in ( 141 ), it has to be assumed that the deletion operation is specific to v erb doubling constructions. Indeed, Lai ( 2019 ) pr oposes a similar deletion operation in his anal y sis on dislocation cop ying, and suggests that a deletion operation specificall y app lies to dislocation cop ying. H o w ev er , it is unclear w h y such a construction-specific mechanism has to be adopted, especiall y w hen ther e is a a mor e straightf orw ar d, head-to-specifier mo v ement anal y sis. I ther ef or e conclude that the pr oposed head mo v ement to specifier is superior to the alternativ es discussed in these tw o subsections, primaril y because its app lication does not r el y on other mecha- nisms such as scrambling and deletion. 111 3.8. D ISCUSSI O NS AND IMP LI CA TI O NS 3.8 Discussi ons and imp li cati ons 3.8.1 R ef ormulating the H ead M o v ement Constr aint If the pr oposed head mo v ement anal y sis is on the right track, then v erb doubling constructions in Cantonese constitute empirical evidence against the H ead M o v ement Constraint (HM C), since v erb mo v ement can skip interv ening heads ( alongside the examp les discussed in § 3.2 ). In other w or ds, the HM C cannot be maintained as a general locality condition on head mo v ement. Indeed, it has been pr oposed suggested in passing that the locality constraint on head mo v ement should be “ attributed to the pr esence of r elev ant f eatur es on the interv ening head( s )” (T o y oshima 2001 , p.121). Lik ewise, R oberts ( 2001 ) also mentions in a f ootnote that “ a f eatural characterization of the interv ener f or head mo v ement seems mor e than justified in this context” (p.147, fn11). If so, then the evidence pr eviousl y tak en to support the HM C is no w in need of exp lanation. The question becomes w h y head mo v ement is often short/local in V - T / T -C mo v ement in R omance and Germanic languages, f or examp le. T o deriv e the local natur e of head mo v ement, one possibility is to r esort to categorial selection, or C-selection , as suggested in M atushansky ( 2006 ). (144) Categorial selection ( C-selection, M atushansky 2006 , p.76) A head ma y select the syntactic category ( and the lexical content ) of the head of its comp lement. As pr oposed in S v enonius ( 1994 ), H olmberg ( 2000 ), and J ulien ( 2002 ), C-selection is achiev ed via a set of C- f eatur es, w hich can be construed as ( uninterpr etable ) counterparts of categorial f eatur es. Cru - ciall y , M atushansky ( 2006 ) suggests that head mo v ement is based on C-selection and the local natur e of head mo v ement is not due to head mo v ement per se , but it is a dir ect consequence of the local na- tur e of C-selection. F or examp le, a T head has an uninterpr etable [ u V] f eatur e and C-selects a VP as its comp lement. V erbs, with the interpr etable [ i V] f eatur e, mo v e to T to check the f eatur e on T . 67 67. This pr o vides a partial exp lanation of w h y man y cases of head mo v ement seem to be semanticall y inert (Chomsky 2001 , p.37). This is because C-selection in v ol v es f ormal f eatur es, and thus it does not trigger interpr etiv e eff ects. 112 3.8. D ISCUSSI O NS AND IMP LI CA TI O NS (145) Illusions of head mo v ement trigger ed b y C-selection/ C- f eatur es a. CP C TP T [ u V] VP V [ i V] b. CP C [ u T] TP T [ i T] VP V In other w or ds, head mo v ement is local because the attracting C- f eatur e r esides in the next higher head in the structur e, but not because head mo v ement cannot skip an interv ening head. This exp lanation to the local natur e of head mo v ement allo w s head mo v ement to be non- local, if it is not trigger ed b y C- f eatur e. This is pr ecisel y the cases of v erb doubling constructions in Can- tonese. As pr oposed, v erb mo v ement is trigger ed b y the [ u F ocus ]/[ u Def ocus ] f eatur es, w hich occur at a distance fr om the v erb (i.e., in the CP domain ). V erb mo v ement into the specifier position of these pr ojections ma y skip heads along its mo v ement path. H o w ev er , it cannot skip an element that bears an identical [ i F ocus ] f eatur e. This locality r equir ement ma y be tak en to motiv ate a f eatural r ef ormulation of the HM C, w hich states that head mo v ement cannot skip interv ening ( matching) f eatur es , instead of heads . 68 M or e generall y , the locality r equir ement on head mo v ement can be subsumed under the locality condition of Agr ee, as pr oposed in § 3.6 . Pr o vided that p hrasal mo v ement is also subject to the locality condition of Agr ee, a non-trivial imp lication of this suggestion is that head mo v ement is constrained in a w a y similar to p hrasal mo v ement, and ther e is no head- mo v ement -specific constraint such as HM C. Bef or e I lea v e this subsection, I briefl y discuss a comp lication in the suggestion that local head mo v ement is connecte d to C-select ion ( or an y f eatural trigger ). This issue has been discussed in M a- tushansky ( 2006 ), F unak oshi ( 2014 ), and Pr eminger ( 2019 ). In the structur e in ( 145 ), it is p lausible that VP also bears the same [ i V] f eatur e as V , w hich in turns constitutes a structurall y closer goal f or [ u V] on T . Consequentl y , VP should mo v e instead of V . This pr oblem can be generalized to an y p hrasal 68. A similar idea has been suggested in R oberts ( 2001 ), w her e Long H ead M o v ement in B r eton can mo v e acr oss non- operator heads such as auxiliaries but not operator heads such as negation. 113 3.8. D ISCUSSI O NS AND IMP LI CA TI O NS category XP , as it will block the mo v ement of its o wn head X. As such, without further assumptions, deriving the local natur e via C-selection undesirabl y rules out all instances of local head mo v ement. Diff er ent pr oposals ha v e been made to dismiss the status of XP as an interv ener of the mo v ement of its o wn head X. F or examp le, F unak oshi ( 2014 ) suggests that in the structur e in ( 145 ), VP does not serv e as an interv ener because it cannot mo v e to the specifier of TP , as it is too “local” (based on notion of anti-locality (Abels 2003 ; Gr ohmann 2003 )). Alternativ el y , Pr eminger ( 2019 ) suggests that VP ceases to interv ene due to a prior Agr ee r elation with the T head. M o v ement of V out of VP does not violate locality constraints ( specificall y minimality constraints ), since locality constraints onl y need to be satisfied once (f ollo wing the spirit of Princip le of Minimal Comp liance, or P M C, N . Richar ds ( 1998 )). T o the extent that anti - locality and the P M C ar e independentl y motiv ated, the v alidity of these exp lanations hinges on the corr esponding pr edictions on head mo v ement, w hich I do not dw ell on her e. 3.8.2 A par all el anal y sis with p hr asal mo v ement In addition to eliminating an asymmetry in locality constraints betw een head mo v ement and p hrasal mo v ement, the mo v ement anal y sis of the f our v erb doubling constructions further suggest that the r elev ant constructions do not distinguish their targets based on structural types (i.e., heads and p hrases ). R ecall that the examp les with a disp laced object in ( 27 ), and consider also the examp les in ( 146 ) with a d isp laced VP . 114 3.8. D ISCUSSI O NS AND IMP LI CA TI O NS (146) The constructions that can target v erbs can also target v erb p hrases a. Z igei self zyu cook faan rice sik eat Aaming Aaming hai co p m- hang not - willing ge2. sfp Daan but k eoi he hang willing bongsau help sai w ash wun. dish ‘Cooking on his o wn, Aaming is not willing to ( do so ). But he is willing to help w ash dishes. ’ b. Lin ev en dim - haa touch-d el ni -zek this-cl dungmat animal Aaming Aaming dou also m-gam. not -dar e ‘ Aaming does not ev en dar e to T O U CH THIS ANIMAL. ’ c. H ai co p heoi go haangsaan hiking Aaming Aaming m-soeng not - w ant zaa3. sfp Zou do k eita other je thing k eoi he wui will heoi. go ‘ Aaming does not w ant to GO HIKIN G onl y . H e will do other things. ’ d. Aaming Aaming kyutding - zo decide--perf gaa3 sfp laa3 sfp heoi go M eigw o k US duks yu . study ‘ Aaming has decided to go to the US f or study . ’ These cases suggest that the each of these constructions can target both heads ( v erbs ) and p hrases ( nominal objects and v erb p hrases ). The onl y diff er ence concerns the size of the f ocused/ def ocused elements ( and the doubling eff ects ). An updated pattern including the VP cases is giv en in T able 3.12 . (i) topic (ii) ‘ ev en ’ - f ocus (iii) copula f ocus (iv ) right dislocation Object O S V lin - O S V hai - O S V S V sfp O V erb V S V O lin - V S V O hai - V S V O S V O sfp V VP VP S V lin - VP S V hai - VP S V S V sfp VP T able 3 .12: The w or d or der patterns illustrate d in ( 27 ), ( 28 ) and ( 146 ) 115 3.8. D ISCUSSI O NS AND IMP LI CA TI O NS The parallels f ollo w naturall y fr om the pr oposal. As I str essed in the pr oposal in § 3.6 , the f o- cus/ def ocus mo v ement is due to the pr esence of an [ i F ocus ]/[ i Def ocus ] f eatur e on constituents, w hich can be a head or a p hrasal element. In other w or ds, the mo v ement under discussion is not designed specificall y f or v erb mo v ement; instead, it r eadil y accounts f or mo v ement other v erbs, if the r elev ant f eatur es ar e associated with p hrasal elements. 69 This idea is schematicall y r epr esented in ( 147 ), sho w - ing the r elev ant structur e bef or e merging with sentence- final particles ( and the SFP -driv en mo v e- ment ). (147) A unif orm mo v ement anal y sis a. [ i F ocus ]/[ i Def ocus ] on v erbs, as in ( 28 ) ... [ F ocus P /Def ocus P Spec F ocus/Def ocus ... [ VP V [ i F ocus/ i Def ocus ] Object ] ] b. [ i F ocus ]/[ i Def ocus ] on v erb p hrases, as in ( 146 ) ... [ F ocus P /Def ocus P Spec F ocus/Def ocus ... [ VP [ i F ocus/ i Def ocus ] V Object ] ] c. [ i F ocus ]/[ i Def ocus ] on objects, as in ( 27 ) ... [ F ocus P /Def ocus P Spec F ocus/Def ocus ... [ VP V Object [ i F ocus/ i Def ocus ] ] ] If all these cases ar e deriv ed unif orml y via Agr ee on the F ocus/Def ocus f eatur e, f ollo w ed b y subse- quent mo v ement, a pr ediction is that the VP mo v ement cases and the object mo v ement cases exhibit F ocus Interv ention Eff ects, in the same w a y as v erb doubling constructions. I discuss this issue in the next subsection. 69. I sland eff ects of constructions lik e ( 27 ) ar e r eported in literatur e, w hich I do not r epeat her e. F or topic constructions, see H uang, Li, and Li ( 2009 ), f or an o v erview; f or ‘ ev en ’ - f ocus constructions, see Sh yu ( 1995 ); f or copula f ocus construc - tions, see C. C.-H. Cheung ( 2008 , 2015 ); f or right dislocation/ dislocation cop ying, see L. Y .-L. Cheung ( 2015 ) and Lee ( 2017 ). 116 3.8. D ISCUSSI O NS AND IMP LI CA TI O NS 3.8.3 F ocus Interv enti on Eff ects in p hr asal mo v ement The pr ediction on F ocus Interv ention Eff ects in the p hrasal mo v ement cases ar e onl y borne out par - tiall y . F irst, in the VP , F ocus Interv ention Eff ects ar e observ ed in topic constructions, ‘ ev en ’ - f ocus constructions and copula f ocus constructions, to the ex clusion of dislocation cop ying. 70 All the sen- tences in ( 148 ) contain a dak -associate in the subject position. (148) F ocus Interv ention Eff ects observ ed with v erb p hrases a. * Z igei self zyu cook faan rice sik , eat dak onl y Aaming Aaming hai co p m- hang not - willing ge2. sfp ‘Cooking on his o wn, onl y Aaming is not willing to ( do so ). ’ b. ?? Lin ev en dim - haa touch-d el ni -zek this-cl dungmat animal ngo I dou also gok dak think dak onl y Aaming Aaming m-gam. not -dar e ‘I think that Aaming does not ev en dar e to T O U CH THIS ANIMAL. ’ c. * H ai co p heoi go haangsaan hiking dak onl y Aaming Aaming m-soeng not - w ant zaa3. sfp ‘Onl y Aaming does not w ant to GO HIKIN G onl y . ’ d. Dak onl y Aaming Aaming kyutding - zo decide--perf zaa3 sfp heoi go M eigw o k US duks yu . study ‘ Aaming has decided to go to the US f or study . ’ These cases ar e consistent with the pr oposed mo v ement account, and F ocus Interv ention Eff ects ar e observ ed in both v erb doubling constructions and their p hrasal counterparts. H o w ev er , in the object cases, no F ocus Interv ention Eff ects ar e observ ed. The sentences in ( 149 ) contain either a wh -expr ession or a dak -associate in the subject position, and they ar e all acceptable. 70. The absence of F ocus Interv ention Eff ects is expected f or ( 148 d), since the mo v ement in right dislocation does not in v ol v e a [F ocus ] f eatur e. 117 3.8. D ISCUSSI O NS AND IMP LI CA TI O NS The sentences in ( 149 a-c ) thus posit a challenge to a f ocus mo v ement anal y sis. 71 (149) N o F ocus Interv ention Eff ects observ ed with objects a. N i -bun this-cl s yu book bingo/ w ho dak onl y Aaming Aaming maai - zo. buy -perf ‘W ho b ought this book?/ Onl y Aaming bought this book. ’ b. Lin ev en ni -bun this-cl s yu book ngo I dou also jingw ai think [ dak onl y Aaming Aaming tai - zo ]. r ead-perf ‘E v en f or this book, I think onl y Aaming ha v e r ead it. ’ c. H ai co p ni -gaan this-cl uk house dak onl y Aaming Aaming zou - m- hei r ent - not -up ze1. sfp ‘I t is onl y this house that onl y Aaming cannot aff or d r enting. ’ d. dak onl y Aaming Aaming m- zi not - kno w zaa3 sfp ni - joeng this-cl je . thing ‘Onl y Aaming doesn ’t kno w this. ’ I suggest that the absence of F ocus Interv ention Eff ects in sentences in ( 149 ) is due to the fact that these sentences ha v e a non- mo v ement deriv ation. In particular , I suggest that the disp laced objects in these cases ar e base generated in the Spec F ocus P and they ar e co- index ed with a null pr onominal element ( e.g., a pr o , w hich in common in Chinese languages ) in the base position. The possibility of a base generation appr oach in constructions lik e the ones in ( 148 a-c ) has been suggested in the literatur e ( see the r ef er ences in f ootnote 69 ), and is illustrated in the “ gap less” structur es in ( 135 ). As such, the structur e in ( 150 ) does not in v ol v e mo v ement, and the f ocused subject does not lead to F ocus Interv ention Eff ects. 71. Again, the acceptability of ( 149 d) is expected, as it d oes not in v ol v e a [F ocus ] f eatur e. 118 3.9. CO N CL USI O NS (150 ) A base-generation appr oach to sentences lik e ( 149 a-c ) [ F ocus P Object i [ i F ocus ] F ocus ... [ TP S [ i F ocus ] V pr o i ] ] I t should be noted that the a v ailability of a non- mo v ement deriv ation is contingent on the a v ail - ability of the appr opriate null elements. The r eason w h y cases of v erb doubling constructions and the sentences in ( 148 ) cannot cir cum v ent F ocus Interv ention Eff ects in a similar w a y can be attributed to the fact that ther e is no corr esponding v erbal pr o elements. In other w or ds, the asymmetry betw een v erb doubling constructions and ( 148 ) on one hand ( 149 ) on the other is due to the lexical r esour ces of null elements in Cantonese. If the discussion her e is on the right track, then the v erb doubling constructions can be anal yzed on a par with their p hrasal counterparts under the curr ent mo v ement -to-specifier pr oposal, and F ocus Interv ention Eff ects, as a r esult of the locality condition of Agr ee, ar e observ ed with both heads and p hrasal elements in constructions that in v ol v e f ocus mo v ement. 3.9 Conclusi ons This chapter examined potential interv ening elements in head mo v ement b y in v estigating f our cases of non- local v erb disp lacement in Cantonese. In these cases, the v erbs ar e doubled, and their cop y appears in the initial or final position of the sentence. I pr oposed that these f our cases unif orml y in v ol v e head mo v ement to a specifier position in the CP perip hery , in a w a y identical to their p hrasal counterparts. I further argued that elements of the same structural types (i.e., heads/ v erbs ) do not necessaril y block the pr oposed mo v ement; instead, elements that possess the same syntactic f eatur e ar e genuine interv eners. The findings in the chapter challenge the status of the H ead M o v ement Constraint as a general constraint on head mo v ement. A t the same time, I sho w that the pr oposed head mo v ement exhibits the syntactic interv ention eff ects that ar e commonl y observ ed with p hrasal mo v ement. I conclude that head mo v ement is not constrained in a w a y diff er ent fr om p hrasal mo v ement with 119 3.9. CO N CL USI O NS r egar d to interv ention. P articularl y , interv ention eff ects ar e calculated in terms of syntactic f eatur es but not structural types. This conclusion necessitates a mo v ement theory that does not distinguish head mo v ement fr om p hrasal mo v ement in terms of locality . 120 Chapter 4 Scope eff ects: mo v ement of quantifi cati onal heads Chapter summary: This chapter f ocuses on the semantic eff ects o f head movement, which ar e o ften said to be absent fr om such movement. The primary goal o f this chapter is to pr esent a novel piece o f evidence f or head movement with semantic eff ects fr om Cantonese. An in-depth investiga- tion into the distribution o f quantificational heads such as aspectual verbs and modal verbs in Cantonese shows that these heads can occupy a non-canonical, high position in the sentence, if they ar e immediately f ollowed by a quantificational element or a f ocused element. I pr opose that these quantificational heads can undergo overt head movement to a higher position and take scope in the landing site ( i.e, scope -shifting head movement ). Additionally, the pr oposed movement o f quantificational heads is constr ained by an independently motivated condition on interpr etation, Scope Economy, which pr ecludes semantically vacuous scope -shifting op- er ations. The findings lend support to the claim that head movement can induce semantic eff ects. I mportantly, head movement can shift scope r elations, in a way similar to Quanti- fier R aising pr oposed f or nominal quantifiers. Furthermor e, Scope Economy is shown to be a gener al constr aint on both phr asal movement and head movement. The conclusion o f this chapter challenges the view that head movement does not r esult in any semantic eff ects - a view which has been said to claim that head movement is non-syntactic and occurs at P F . I t is 121 4.1. INTR O D U CTI O N shown that head movement is no diff er ent fr om phr asal movement in the potential to induce semantic eff ects. This motivates a unified theory o f movement o f head movement and phr asal movement. 4.1 Intr oducti on This chapter f ocuses on the debate about semantic eff ects of head mo v ement, w hich ar e often said to be absent (Chomsky 2001 ; H arley 2004 , 2013 ; Platzack 2013 ). S uch a lack of semantic eff ects appears to distinguish head mo v ement fr om p hrasal mo v ement. This raises non-trivial concerns r elating to the theor etical status of head mo v ement in mo v ement theories. A primary goal of this chapter is to pr esent a no v el piece of evidence f or head mo v ement with semantic eff ects fr om Cantonese. The cor e empirical f oundations come fr om the v ariable distribution of aspectual v erbs and modal v erbs in Cantonese. F or examp le, the aspectual v erb hoici ‘begin ’ can appear in either a post -subject (lo w ) position or a pr e-subject (high ) position, as illustrated in ( 151 ). I t is significant to note that he diff er ent surface positions of hoici in ( 151a ) and ( 151b ) corr espond to diff er ent interpr etations with r egar d to the subject mark ed b y the f ocus mark er dak ‘ onl y’ . In both cases, onl y the surface scope r eading is a v ailable. (151) The lo w and high positions of hoici ‘begin ’ a. ‘ onl y’ > ‘begin ’ / *‘begin ’ > ‘ onl y’ Dak onl y Aaming Aaming hoi ci begin haau -dou get -able hou good singzik. r esult ‘Onl y Aaming is such that he begins to get good r esults. ’ b. *‘ onl y’ > ‘begin ’ / ‘begin ’ > ‘ onl y’ H oi ci begin dak onl y Aaming Aaming haau -dou get -able hou good singzik. r esult ‘I t begins to be the case that onl y Aaming is getting go od r esults. ’ Importantl y , the p lacement of hoici ‘begin ’ is not unconstrained. In the absence of the f ocus par - 122 4.1. INTR O D U CTI O N ticle dak , the high, pr e-subject position is una v ailable, as in ( 152b ). (152) The high position of hoici ‘begin ’ una v ailable in the absence of dak ‘ onl y’ a. Aaming Aaming hoi ci begin haau -dou get -able hou good singzik. r esult ‘ Aaming begins to get good r esults. ’ b. * H oi ci begin Aaming Aaming haau -dou get -able hou good singzik. r esult Int: ‘ Aaming begins to get good r esults. ’ S imilar patterns ar e also observ ed with modal v erbs lik e hoji ‘ma y’ and wui ‘will’ . The distribution of aspectual v erbs and modal v erbs and the r estriction on the high position calls f or an anal y sis of the paradigm in ( 151 ) and ( 152 ). As far as I kno w , this has not been documented in the Cantonese literatur e. 1 Empiricall y , this chapter mak es the no v el observ ation that the distribution of aspectual v erbs and modal v erbs in Cantonese is corr elated with the pr esence of quantificational and f ocused elements in the pr e- v erbal position. Anal yticall y , I pr opose that these v erbs can undergo o v ert head mo v ement to a high position and tak e scope in the landing site. In other w or ds, they can undergo scope-shifting head mo v ement. A dditionall y , I suggest that the pr oposed mo v ement is constrained b y an indepen- dentl y motiv ated condition on interpr etation, namel y , Scope Econom y , w hich pr ecludes semanticall y v acuous scope-shifting operations (F o x 2000 ). The findings lend support to claims r ecentl y f ound in the literatur e that head mo v ement can im- pose semantic eff ects. Importantl y , head mo v ement can shift scope r elations, in a w a y similar to Quantifier Raising pr oposed f or nominal quantifiers. F urthermor e, Scope Econom y i s sho wn to be a 1. The structural position of modal v erbs in M andarin is not a new topic, but the discussions f ocus on the a v ailabil - ity of the pr e-subject position of epistemic modals, in contrast to other modal v erbs (Lin 2011 ; Chou 2013 ; T sai 2015 ). One ex ception is Y .-y . H su ( 2016 , 2019 ), w ho discusses admissible cases of pr e-subject deontic modals. But ther e is no discussion on the positional alternation of aspectual v erbs, as far as I am a w ar e. 123 4.2. (N O N -)OCCURREN CE O F SEMANTI C EFFECTS WITH HEAD M O VEMENT : AN O N GO IN G D EB A TE general constraint on both p hrasal mo v ement and head mo v ement. M or e generall y , t he conclusion of this chapter challenges the assumption that a lack of semantic eff ects is a general empirical pr operty of head mo v ement, a pr operty that is tak en as evidence f or the non-syntactic status of head mo v ement ( an anal ytical position f ound in Chomsky 2001 ; H arley 2004 , 2013 ; Boeckx and S tjepano vić 2001 ; Schoorlemmer and T emmerman 2012 ; Platzack 2013 ; H all 2015 ; M c Closk ey 2016 ). I t is sho wn that head mo v ement is no diff er ent fr om p hrasal mo v ement in its po- tential to impose semantic eff ects. F urthermor e, the observ ation that head mo v ement is constrained b y an econom y condition that w as originall y pr oposed to constrain ( some instances of ) p hrasal mo v e- ment (in particular Quantifier Raising) pr o vides new evidence f or the claim that the computational sy stem of natural languages does not discriminate head mo v ement fr om p hrasal mo v ement. This sets out the f oundations of a unified theory of mo v ement. The organization of this chapter is as f ollo w s. In § 4.2 , I r eview the debate about semantic eff ects with head mo v ement. In § 4.3 , I in v estigate the distribution of aspectual v erbs and modal v erbs in Can- tonese. In § 4.4 , I detail the pr oposed head mo v ement anal y sis, and, in § 4.5 , I pr o vide further evidence f or the pr oposed anal y sis. In § 4.6 , I discuss and argue against alternativ e anal y ses to a head mo v ement appr oach. In § 4.7 , I discuss some consequences and imp lications of the pr oposal. I conclude in § 4.8 . 4.2 (N on -)occurr ence of semanti c eff ects with head mo v ement: an ongoing debate The debate about semantic eff ects with head mo v ement ( w hether such eff ects occur or not ) is part of the br oader debate about the theor etical status of head mo v ement in the generativ e/ minimalist literatur e. S ince the theor etical r ecognition of head mo v ement as a syntactic operation in the gram- mar (K oopman 1984 ; T ra vis 1984 ; Bak er 1985 , 1988 ), its diff er ences fr om other (p hrasal) mo v ement operations such as A - mo v ement ( mo v ement to argument positions ) and A ’ - mo v ement ( mo v ement to non-argument positions ) ha v e supported a non-unif orm anal y sis of head/ p hrasal mo v ement depen- 124 4.2. (N O N -)OCCURREN CE O F SEMANTI C EFFECTS WITH HEAD M O VEMENT : AN O N GO IN G D EB A TE dencies. Among other diff er ences, head mo v ement is often said to lack semantic eff ects, as opposed to ( m an y ) instances of p hrasal mo v ement (Chomsky 2001 ; H arley 2013 ; Platzack 2013 ). This has stim- ulated a debate on w hether the appar ent lack of semantic eff ects with head mo v ement is a general empirical pr operty of the mo v ement of heads. The significance of the debate is that if head mo v ement can impose semantic eff ects, this w ould constitute evidence against a non-syntactic anal y sis of head mo v ement. 2 In fact, if it turns out that head mo v ement cannot impose an y semantic eff ects, tr eating head mo v ement as a syntactic operation needs to dev elop a p lausible account of w h y ther e should be this diff er ence with p hrasal mo v ement ( w hich can impose semantic eff ects ). A dditionall y , the debate hinges on w hether head mo v ement should in general be distinguished fr om p hrasal mo v ement, since ther e is gr o wing evidence that the occurr ence of semantic eff ects does appear to cr osscut both head mo v ement and p hrasal mo v ement. In § 4.2.1 , I r eview the evidence in support of assumptions that ther e is a lack of semantic eff ects with head mo v ement. In § 4.2.2 and § 4.2.3 , r espectiv el y , I r eview the debates and concerns on the arguments f or the semantic eff ects of head mo v ement based on (i) discourse eff ects and (ii) scope ef - f ects. 3 4.2.1 A lack of semanti c eff ects? The primary observ ation of an appar ent lack of semantic eff ects with head mo v ement can be at - tributed to v erb mo v ement in Germanic and R omance languages ( cf. Emonds 1978 ; P ollock 1989 ). F or examp le, H arley ( 2004 ) notes that v erb mo v ement o v er negation in F r ench does not cr eate diff er - ent scope r eadings betw een the v erb and the negation (i.e., V > NEG/ NEG > V ). In both sentences in 2. W hile it will not be the f ocus of this c hapter , evidence f or the syntactic natur e of head mo v ement also includes v arious syntactic eff ects of head mo v ement. F or examp le, head mo v ement ma y license ellipsis (Gergel 2009 ); it ma y r ef or - mulate/ v oid opaque syntactic domains (den Dikk en 2006 ; Gallego 2010 ; S tepano v 2012 ); it ma y f eed subsequent syntactic mo v ement such as VP - fr onting (W iland 2008 ; F unak oshi 2014 , 2019 ). 3. I ha v e set aside a f ew studies, such as Benedicto ( 1998 ), Lechner ( 1998 ), Z w art ( 2001 ), H artman ( 2011 ), K eine and B hatt ( 2016 ), and Gribano v a ( 2017 ) w ho discuss mo v ement of heads that imposes semantic eff ects other than discourse eff ects and scope eff ects. F or discussions of some of these studies, see H all ( 2015 ) and M c Closk ey ( 2016 ). 125 4.2. (N O N -)OCCURREN CE O F SEMANTI C EFFECTS WITH HEAD M O VEMENT : AN O N GO IN G D EB A TE ( 153 ), the negation tak es scope o v er the v erb. 4 (153) F r ench V mo v ement to T (H arley 2004 , p.244) a. The main verb moves over negation NEG > V J e an J e an ne ne par lait speak.imp pas not français. F r ench ‘J o hn di dn ’t speak F r ench. ’ b. The m ain verb does not move over negation NEG > V J ean J ean n ’ as has pas not par l é spok en français. F r ench. ‘J o hn ha sn ’t spok en F r ench. ’ H arley ( 2013 ) also suggests that, in English, the position of the past tense morp heme does not aff ect the r elativ e scope of tense and the univ ersal quantifier in case of T -C mo v ement, as sho wn in ( 154 ). Both sentences can ha v e wide and narr o w scope r eadings f or the time v ariables intr oduced b y tense. (154) (H arley 2013 , p.117) English V /T mo v ement mo v ement to C a. E v ery one l eft . ev ery one > p ast; p ast > ev ery one b. Di d ev ery one lea v e? ev ery one > p ast; p ast > ev ery one A dditionall y , Platzack ( 2013 ) examines minimal pairs of examp les in diff er ent languages, w her e v erb mo v ement is a v ailable in one l anguage but una v ailable in the other . As she suggests, if v erb/head 4. In I talian, C inque ( 1999 , p.184, fn.8) mentions that the diff er ent positions of past particip le with r egar d to adv erbs lik e sempr e ‘ al w a y s’ ma y ha v e diff er ent interpr etations. W hile the “ al w a y s > particip le” or der onl y has a non-perf ect r eading, the “particip le > al w a y s” or der is compatible with both a perf ect and non-perf ect r eading. (i) Placement of particip les in I talian a. Gianni ha sempr e a vuto i capelli l unghi. ‘Gianni still has long hair . ’ a non-perf ect r eading b. Gianni ha a vuto sempr e i capelli l unghi. ‘Gianni had/ still has long hair’ a perf ect or a non-perf ect r eading 126 4.2. (N O N -)OCCURREN CE O F SEMANTI C EFFECTS WITH HEAD M O VEMENT : AN O N GO IN G D EB A TE mo v ement has semantic eff ects, it is expected to cause diff er ent r eadings. H o w ev er , this does not appear to be the case. F or examp le, English lacks v erb mo v ement to the second position, as opposed to S w edish. H o w ev er , the tw o sentences in ( 155 ) ar e semanticall y e quiv alent. (155) (Platzack 2013 , p.31) Languages with/ without v erb mo v ement to the second position a. (English, a non- V2 language ) This book, J ohn has r ead. b. ( S w edish, a V2 language ) Denna this bok book har has J ohn J ohn läst. r ead ‘This book, J ohn has r ead. ’ T o see one mor e examp le, Platzack ( 2013 , p.32-33) argues that V2 mo v ement of N egativ e P olarity V erbs in N orw egian and S w edish does not alter the c -commanding r elation with the negativ e licensor . F or examp le, in ( 156 a ), the v erb enset ‘notice’ in N orw egian r equir es negativ e licensing ( as the absence of negati on leads to unacceptabil ity ). Assuming that the V2 position is higher than the negation ikke ‘not’ , V2 mo v ement of enset ‘notice’ is f ound not to aff ect the negativ e licensing of enset . This suggests that V2 mo v ement does not impose syntactic/ semantic eff ects. S imilar conclusions can be dra wn fr om the V2 mo v ement of gitter ‘bother’ in S w edish in ( 156 b ). (156) V2 mo v ement of N egativ e P olarity V erbs a. (N orw egian, Platzack 2013 , p.32) H un she enset sensed *(ikk e ) not bråk et. noise.def ‘She didn ’t notice the noise. ’ b. ( S w edish, Platzack 2013 , p.33) H an he gitter bothers *(inte ) not göra to.do det. it ‘H e doesn ’t bother to do it. ’ These cases of head mo v ement substantiate the suggestion in Chomsky ( 2001 , p.37) that: “ seman- tic eff ects of head raising in the cor e inflectional sy stem ar e slight or nonexistent, as contrasted with 127 4.2. (N O N -)OCCURREN CE O F SEMANTI C EFFECTS WITH HEAD M O VEMENT : AN O N GO IN G D EB A TE XP - mo v ement, with eff ects that ar e substantial and sy stematic. ” In other w or ds, the lack of semantic eff ects seem to be characteristic, empirical pr operty of head mo v ement. 5 H o w ev er , all these cases concern v erb mo v ement in Germanic and R omance languages. This raises questions of generality , as head mo v ement has been pr oposed f or v arious p henomena in diff er ent lan- guages. Also, ther e ar e at least tw o r easons f or the consistent lack of semantic eff ects of in these cases. 6 The first one concerns the trigger of the mo v ement. If a mo v ement is trigger ed b y categorial f eatur es (S v enonius 1994 ; H olmberg 2000 ; J ulien 2002 ), such as a V f eatur e on T heads (in V - T mo v ement ) or a T f eatur e on C heads (in T -C mo v ement ), then the mo v ement might not be expected to impose semantic eff ects, since categorial f eatur es ar e commonl y assumed to be pur el y f ormal/ syntactic. An- other r eason, as suggested in M atushansky ( 2006 ) and V icente ( 2007 ), concerns the semantic types of the mo ving heads. The most discussed cases of head mo v ement in v ol v e non-quantificational heads. Pr esumabl y , they ar e of semantic type <e,t> or <e,<e,t>>. Their interpr etation should be the same in the launching site or the landing position. These suggestions pr o vide an exp lanation on w h y the cases in ( 153 ) thr ough ( 156 ) lack semantic eff ects. Importantl y , these suggestions also mak e a pr ediction on the a v ailability of semantic eff ects of head mo v ement, as described n ( 157 ). (157) H ead mo v ement ma y impose semantic eff ects if a. the mo v ement is trigger ed b y f eatur es other than categorial/ pur el y f ormal f eatur es; or b. the head is of a quantificational type (i.e., << ,t>,t>). I t is thus an empirical question as to w hether such cases of head mo v ement ar e attested in natural language. I r eview certain r elev ant cases in the next tw o subsections. 5. Chomsky ( 2001 , p.37) sets aside cases of noun incorporation, w hich arguabl y in v ol v es head mo v ement (Bak er 1988 ). H arley ( 2013 , p.117, fn.3) suggests that the semantic eff ects of noun incorporation “ ar e not intr oduced b y head mo v ement, but rather a pr econdition on its occurr ence. ” S ince ther e ar e also debates on w hether noun incorporation in v ol v es head mo v ement ( e.g., Barrie and M athieu 2016 ), I set aside these cases and f ocus on cases of v erb mo v ement. 6. The lack of semantic eff ects is often tak en to be evidence f or a p honological appr oach to head disp lacement (Chom- sky 2001 ; H arley 2004 , 2013 ; Platzack 2013 ). H o w ev er , as Platzack ( 2013 , p.34) also admits, p hrasal mo v ement ma y also lack semantic eff ects. Thus t he lack of semantic eff ects does not necessitate a p honological appr oach to head disp lacement. 128 4.2. (N O N -)OCCURREN CE O F SEMANTI C EFFECTS WITH HEAD M O VEMENT : AN O N GO IN G D EB A TE 4.2.2 Discour se eff ects of head mo v ement Concerning ( 157 a ), it is indeed not uncommon to see v erb mo v ement being associated with discourse eff ects. F or examp le, a v erb ma y be interpr eted as a topic or a f ocus in pr edicate cleft/ v erbal fr onting constructions (V icente 2007 ; Cheng and V icente 2013 ; H ein 2018 ; H arizano v 2019 , among man y others; see also r ef er ences in Chapter 3). The v erb doubling constructions in Cantonese discussed in Chapter 3 also fall into this category . A dditionall y , it has been pr oposed that v erb mo v ement to the second position/ the C position imposes illocutionary eff ects, e.g., declarativ e f or ces and interr ogativ e f or ces (W echsler 1991 ; T ruck enbr odt 2006 ). H o w ev er , H arizano v and Gribano v a ( 2019 ) suggests that the discourse eff ects accompan ying mo v e- ment operations ma y be ( at least as a logical possibility ) attributed to the f eatural encoding on the heads that trigger the mo v ement, instead of the mo v ement itself. These heads ma y be a topic head or a f ocus head in the CP domain bearing a [ topic ] or [ f ocus ] f eatur e that triggers v erb mo v ement. Lik ewise, the illocutionary eff ects in V2 languages ar e suggested to be due to the trigger of the head mo v ement, instead of a consequence of head mo v ement (W echsler 1991 ; T ruck enbr odt 2006 ). 7 A ccor dingl y , head mo v ement with discourse eff ects or illocutionary eff ects ma y not serv e as a knock -do wn argument f or the existence of head mo v ement with semantic eff ects. 8 This shifts the spotlight onto cases r elating to ( 157 b ), since head mo v ement with scope eff ects is immune to alternativ e exp lanations of the type just described. The scope eff ects must acco r dingl y be attributed to head mo v ement per se , instead of the trigger . 4.2.3 Scope eff ects of head mo v ement W ith r egar d to ( 157 b ), ther e is a gr o wing body of evidence f or the scope eff ects of head mo v ement, in v ol ving the mo v ement of determiners, negation, modal v erbs and aspectual v erbs (T akahashi 2002 ; 7. Indeed, W iklund ( 2010 ) argues that the illocutionary eff ects ar e not due to V2 mo v ement. 8. H o w ev er , if this r easoning goes thr ough f or head mo v ement, it should also app l y to p hrasal mo v ement that is trig - ger ed b y a topic or f ocus f eatur e. This amounts to the suggestion that man y instances of p hrasal mo v ement lack semantic eff ects in the same w a y as head mo v ement. 129 4.2. (N O N -)OCCURREN CE O F SEMANTI C EFFECTS WITH HEAD M O VEMENT : AN O N GO IN G D EB A TE H an, Lidz, and M usolino 2007 ; Lechner 2007 ; Kishimoto 2007 ; R oberts 2010 ; S zabolcsi 2010 , 2011 ; I atridou and Zeijlstra 2013 ; H omer 2015 ; M atyiku 2017 ; Landau 2020 ; Sato and M aeda 2021 ). H o w - ev er , the r eported evidence f or scope eff ects of head mo v ement is not uncontr o v ersial. Among others, H all ( 2015 ) and M c Closk ey ( 2016 ) criticall y point out that the arguments pr esented ma y build on un- motiv ated assumptions, and/ or ma y ha v e alternativ e anal y ses. F or space r easons, f or each case, I onl y pr esent the cor e observ ations in support of scope eff ects of head mo v ement. I then briefl y mention potential concerns or alternativ e anal y ses discussed in the literatur e. This is meant to illustrate the contr o v ersial natur e of these allege d cases of head mo v ement with scope eff ects. T able 4.1 pr eview s r ecent pr oposals arguing f or the scope eff ects of head mo v ement. H ead Language Scope eff ects R ef er ence( s ) Determiner J apanese enhanced r estriction T akahashi ( 2002 ) N egation English NP I licensing R oberts ( 2010 ) and S zabolcsi ( 2010 ) J apanese NP I licensing Kishimoto ( 2007 ) English v arieties outscope subjects M atyiku ( 2017 ) and Landau ( 2020 ) K or ean outscope objects H an, Lidz, and M usolino ( 2007 ) J apanese outscope objects Sato and M aeda ( 2021 ) M odal v erb English outscope subjects Lechner ( 2007 , 2017 ) English outscope negation I atridou and Zeijlstra ( 2013 ) and H omer ( 2015 ) Aspectual v erb Shupamem outscope subjects S zabolcsi ( 2010 , 2011 ) T able 4.1 : S ummary of evidence of scope eff ects w ith head mo v ement 4.2.3.1 M o v ement of (quantifi cati onal) deter miner s Based on evidence fr om J apanese, T akahashi ( 2002 ) argues that determiner raising ma y enhance the r estriction of quantificational scope. Assuming that the univ ersal mark er mo is a determiner and se- lects wh -expr essions as its comp lement, T akahashi suggests that it ma y head- mo v e to a higher posi - tion. F or examp le, the base generated mo in ( 158 a ) is argued to undergo mo v ement to a higher , NP - external position in ( 158 b ). Cruciall y , mo has an enhanced r estriction in ( 158 b ), i.e., the r estriction of 130 4.2. (N O N -)OCCURREN CE O F SEMANTI C EFFECTS WITH HEAD M O VEMENT : AN O N GO IN G D EB A TE mo is wider if it is adjacent to the matrix subject instead of the embedded subject. (158) (T akahashi 2002 , p.594) J apanese determiner raising a. [ TP [ NP Dar e person mo ]-ga ev ery -n o m kaik osar eru ] is- fir ed to yuu that uw asa - w a rumor -to p hontoo true datta. w as ‘The rumor that ev ery one w ould be fir ed w as true. ’ b. [ TP [ NP Dar e ]-ga person-n o m kaik osar eru ] is- fir ed to yuu that uw asa rumor mo ev ery hontoo true datta. w as Lit.: ‘E v ery rumor that a person w ould be fir ed w as true. This mo v ement anal y sis is supported b y the observ ation that the mo v ement of mo is constrained b y an econom y condition that r equir es its mo v ement to impose semantic eff ects ( Scope Econom y , cf. F o x 2000 ): it must cr oss a quantificational element along its path. Assuming that a noun/ nominal category in v ol v es an imp licit existential determiner , the high position of mo is una v ailable if it onl y cr osses a ( non-quantificational) v erb/ v erbal category , as sho wn in ( 159 b ). (159) (T akahashi 2002 , p.599) J apanese determiner raising, an illicit case a. T ar oo- w a T a r oo-to p [ VP H anak o- ni H anak o-d a t [ NP dar e person mo ]-o ev ery -a cc sikari ] scold sae ev en saseta. made ‘T ar oo made H anak o ev en scold ev ery one. ’ b. * T ar oo- w a T a r oo-to p [ VP H anak o- ni H anak o-d a t [ NP dar e ]-o person-a cc sikari scold mo ] ev ery ( sae ) ( ev en ) saseta. made ‘lit. T ar oo made H anak o ( ev en ) ev ery scold a person. ’ H o w ev er , Y atsushir o ( 2009 , p.167-169) argues that such an appr oach incorr ectl y pr edicts addi - tional positions f or the landing site of mo , w hen ther e is mor e than one nominal category along the mo v ement path. Also, the alleged scope eff ects can be alternativ el y deriv ed without mo v ement under an u nselectiv e binding appr oach, as discussed in Shimo y ama ( 2006 ). 131 4.2. (N O N -)OCCURREN CE O F SEMANTI C EFFECTS WITH HEAD M O VEMENT : AN O N GO IN G D EB A TE 4.2.3.2 M o v ement of negati on Ther e ar e thr ee types of cases concerning the scope eff ects of negation mo v ement in the literatur e, w her e negation interacts with (i) N egativ e P olarity I tems, (ii) subject quantifiers, and (iii) object quan- tifiers. ( i) The licensing scope of negation The first type of cases concerns the licensing scope of negation. Assuming that N egativ e P olarity I tems such as any in English mu st be c -commanded b y an element that licenses it, R oberts ( 2010 , p.8- 12) argues that the mo v ement of the negativ e auxiliary in English ma y extend that scope of negativ e licensing. F or examp le, the mo v ement of doesn’t ( to the C position ) licenses the subject anybody in ( 160 b ). N ote that R oberts argues that n’t optionall y cliticizes to T fr om a lo w er position. W hen T mo v es to C, n’t mo v es together wit h T . (160 ) English T -C mo v ement in question f ormation and NP I licensing a. * [ CP W hich one of them does [ TP an ybody not lik e? ] ] (H arizano v and Gribano v a 2019 , p.514) b. (M c Closk ey 1996 , p.89) [ CP W hich one of them doesn ’t [ TP an ybody lik e ] ] ? S imilarl y , S zabolcsi ( 2010 ) r eports that the mo v ement of negation in imperativ es cr eates an oth- erwise una v ailable licensing context f or subject NP I s, as illustrated in ( 161 ). (161) (S zabolcsi 2010 , p.44) English T -C mo v ement in imperativ es and NP I licensing a. * [ TP An y one of y ou don ’t touch the money ] ! b. [ CP Don ’t [ TP an y one of y ou touch the money ] ] ! H o w ev er , H all ( 2015 ) points out tw o potential concerns with these arguments. On one hand, if n’t cliticizes on T bef or e mo v ement t o C, then it ma y be too embedded to c -command the NP I in the subject position. On the other hand, he suggests that, f ollo wing R oberts’ r ef ormulation of head 132 4.2. (N O N -)OCCURREN CE O F SEMANTI C EFFECTS WITH HEAD M O VEMENT : AN O N GO IN G D EB A TE mo v ement as a special case of Agr ee, the licensing eff ects ma y be due to the Agr ee r elationship betw een the NP I and a negativ e f eatur e w hich is alr eady pr esent in C. 9 If this is the case, T -C mo v ement per se does not contribute to successful licensing. Another case that tak es NP I licensing as evidence f or scope eff ects of head mo v ement is fr om J apanese. The negation - na( i ) in J apanese can license subjects associated with NP I s such as sika ‘ onl y’ . Diff er ent fr om pr evious anal y ses that suggest a lo w position of the NP I subjects, ( e.g., T akahashi 1990 ; A o y agi and I shii 1994 ; Kato 2000 ), Kishimoto ( 2007 ) argues instead that the negation - na( i ) has un- dergone mo v ement to T so that it can license the NP I in subject position. 10 (162) (Kishimoto 2007 , p.264, modified) J apanese negation licensing NP I subjects NEG- T mo v ement Gakusei -sika student -onl y hon-o book -a cc y oma- nakat -ta. r ead-neg-pst ‘Onl y students r ead books. ’ Importantl y , the ability to license NP I subjects disappears if the negation does not undergo mo v e- ment. Kishimoto ( 2007 ) suggests that the negation in the sentence in ( 163 a ) does not mo v e to T , as it is separated fr om T b y the causativ e v erb si ‘mak e’ ( and f ocus particles such as - sae ‘ ev en ’ and - mo ‘ also ’). In such case, it fails to license subject NP I s, as sho wn in ( 163 b ). 11 (163) (Kishimoto 2007 , p.270, modified) J apanese negation without mo ving to T a. J ohn-ga J o hn-n o m M ary -o M ary -a cc hey a- ni r oom-to hair e- naku (-sae/ mo ) enter .can-neg-ev en/ also si -ta. mak e-pst ‘J ohn made M ary unable to enter the r oom. ’ 9. This suggestion is also discussed in H arizano v and Gribano v a ( 2019 , p.513-517). 10. I t is assumed that the w hole TP , instead of the comp lement of the T head, is the licensing domain of negation if negation occupies the T head ( cf. the m-command domain of the T head). 11. Ther e ar e other contexts w her e the negation does not mo v e to T , w hich lik ewise corr elate with the inability to license NP I subjects. F or further discussions, see Kishimoto ( 2007 , 2013 ). 133 4.2. (N O N -)OCCURREN CE O F SEMANTI C EFFECTS WITH HEAD M O VEMENT : AN O N GO IN G D EB A TE b. * J ohn-sika J o hn-onl y M ary -o M ary -a cc hey a- ni r oom-to hair e- naku enter .can-neg si -ta. mak e-pst ‘Onl y J ohn made M ary unable to enter the r oom. ’ H o w ev er , as will be discussed shortl y in sentences lik e ( 167 ), the negation in J apanese cannot outscope objects that ar e mark ed b y dake ‘ onl y , ’ f or examp le. The scope beha viors betw een negation and objects appear to be inconsistent with the negation mo v ement anal y sis adv ocated b y Kishimoto ( 2007 ). ( ii) Scope r elations with subject quantifiers The second type of cases concerns the scope r elation betw een negation and subject quantifiers. M atyiku ( 2017 ) argues at length that in some v arieties of English, such as W est T exas English, African Ameri - can English, and Appalachian English, the auxiliary - first w or d or der , or negativ e auxiliary in v ersion, is deriv ed via a mo v ement of the negated auxiliary to a position higher than the subject. F or examp le, ( 164 b ) is deriv ed fr om ( 164 a ), w her e don’t mo v es o v er many people . S ince the tw o sentences con v ey diff er ent scope r eadings (i.e., many > not v s. not > many ), this constitutes evidence f or the scope eff ects of head mo v ement. 12 (164) (M atyiku 2017 , p.37-38) W est T exas English and negativ e auxiliary in v ersion a. M an y peop le don ’t lik e y ou. b. Don ’t man y peop le lik e y ou. H o w ev er , the tw o concerns posited in H all ( 2015 ) f or arguments dra wing on NP I licensing also 12. Landau ( 2020 ) discusses similar cases in standar d English, w her e the mo v ement of a negativ e auxiliary in question f ormation outscopes subject quantifier , as in (i). (i) N egativ e auxiliary mo v ement enhances scope a. E v erybody didn ’t see the fight. not > ev ery; ev ery > not b. Didn ’t ev erybody see the fight? not > ev ery; *ev ery > not 134 4.2. (N O N -)OCCURREN CE O F SEMANTI C EFFECTS WITH HEAD M O VEMENT : AN O N GO IN G D EB A TE app l y to these cases. M or e importantl y , B lanchette and Collins ( 2019 ) pr opose an alternativ e anal y sis to M atyiku’ s head mo v ement appr oach, w her e the negation is based generated in the subject NP /D P ( e.g., [ S ubj not man y peop le ]). As such, the fact that the negation scopes o v er many people in ( 164 b ) is not because of negation mo v ement, it is argued, but because of diff er ent base generated positions of negation. 13 ( iii) Scope r elations with object quantifiers The thir d type of cases concerns the scope of negation and object quantifiers in languages lik e K or ean and J apanese. H an, Lidz, and M usolino ( 2007 ) r eport that the sentence in ( 165 ) is scopall y ambiguous f or a population of K or ean speak ers. In particular , the ‘NEG > ev ery’ r eading is a v ailable f or some K or ean speak ers they tested, indicated b y the % symbol. (165) (H an, Lidz, and M usolino 2007 , p.24) K or ean short negation and object quantifiers ‘ ev ery’ > NEG; %NEG > ‘ ev ery’ J ohn- i J ohn-n o m motun cha y k -ul ev ery book -a c c an neg ilk -ess ta. r ead-p ast-d ecl a. ‘ ev ery’ > NEG: ‘J ohn r ead no book. ’ b. NEG > ‘ ev ery’: ‘J ohn didn ’t r ead ev ery book. ’ Importantl y , they argue that the a v ailability of such a r eading indicates the mo v ement of negation o v er the object quantifier ‘ ev ery book’ . This argument r elies on thr ee assumptions in the deriv ation of ( 165 ). 14 (166) Assumptions on the deriv ation of the sentence in ( 165 ) a. The ( short ) negation an is adjoined to VP . [ VP an [ VP V Obj. ] ] 13. B lanchette and Collins ( 2019 ) suggests that the negation undergoes negativ e raising fr om the subject position to a higher position, and the auxiliary appearing in the high position is a r esult of do -support f or the contracted negation. 14. F or justifications, see H an, Lidz, and M usolino ( 2007 , p.12-22). 135 4.2. (N O N -)OCCURREN CE O F SEMANTI C EFFECTS WITH HEAD M O VEMENT : AN O N GO IN G D EB A TE b. The object mo v es fr om its base position to a VP -external position. [ FP Obj. [ VP an [ VP V t Obj. ] ] ] c. K or ean is a scope- rigid language (i.e., the scope of a quantificational element is determined b y its surface position ). F or the negation to scope o v er the object quantifier , it must mo v e ( together with the main v erb ) to a position higher than the object quantifier (pr esumabl y a T position ). This constitutes an instance of head mo v ement with scope eff ects. H o w ev er , Zeijlstra ( 2017 ) challenges the generality of the assumption in ( 166 c ): w hile it ma y be true of scope r elations between nominal quantifiers, it does not necessaril y hold true of negation and nominal quantifiers. Object r econstruction belo w the negation is a possible option. 15 On the other hand, H arizano v and Gribano v a ( 2019 ) raise concerns o v er ( 166 a ). They suggest that the speak ers allo wing the ‘N eg > ev ery’ r eading ma y permit flexible positions of an , i.e., one belo w the object and one abo v e the object. If so, no negation mo v ement is needed. Another case tak en as evidence f or scope eff ects of negation is v erb-echo answ ers in J apanese. In J apanese, negation cannot tak e scope o v er objects associated with f ocus-sensitiv e particles such as - dake ‘ onl y’ , as sho wn in ( 167 ). (167) (Shibata 2015 , p.73) J apanese - dake ‘ onl y’ and negation onl y > NEG; *NEG > onl y T ar oo- w a T ar o-to p pan-dak e br ead-onl y ka w - anak -atta. buy -neg-pst ‘T ar o didn ’t buy onl y br ead. ’ Cruciall y , Sato and M aeda ( 2021 ) observ e that the negation ma y scope o v er the dake - mark ed object in v erb-echo answ ers to polarity questions, illustrated in the question-answ er pair in ( 168 ). 15. A similar challenge is pr esented in H arizano v and Gribano v a ( 2019 ) as w ell. 136 4.2. (N O N -)OCCURREN CE O F SEMANTI C EFFECTS WITH HEAD M O VEMENT : AN O N GO IN G D EB A TE (168) (Sato and M aeda 2021 , p.9) J apanese v erb-echo answ ers with negation a. Q: T ar oo- w a T ar o-to p pan-dak e br ead-onl y kat -ta- no? buy -pst-q ‘Did T ar o buy onl y br ead?’ b. ?? ‘ onl y’ > NEG; NEG > ‘ onl y’ A: Ka w a- nakat -ta- y o. buy -neg-pst-sfp Lit.: ‘Didn ’t buy . ’ F ollo wing the clausal ellipsis theory of v erb-echo answ ers pr oposed in H olmberg ( 2016 ), Sato and M a eda ( 2021 ) argue that the deriv ation of v erb-echo answ ers in J apanese in v ol v es V - T -C mo v ement, f ollo w ed b y TP -ellipsis. The wide scope r eading of the negation o v er the dake - mark ed object is a dir ect consequence of negation mo ving together with the v erb in the V - T -C mo v ement. 4.2.3.3 M o v ement of modal v erbs T urning to cases of mo v ement of modal v erbs, Lechner ( 2007 ) pr esents an argument f or scope ef - f ects with head mo v ement based on the scope r elations betw een modal v erbs and negativ e ( univ ersal) quantifiers. The primary observ ation concerns the interpr etation of sentences lik e ( 169 ). (169) (Lechner 2007 , p.3) English modal v erbs and the scope sp litting constructions N ot ev ery bo y can mak e the team. not > ev ery > can ; not > can > ev ery The r eading of inter est is the one w her e the possibility modal v erb can is interpr eted within the scope of negati on but abo v e the univ ersal quantifie r , i.e., not > can > ev ery . This r eading con v ey s that ther e is no possible w orld in w hich all of the bo y s mak es the team. Setting aside man y details of the argu - ment 16 , Lechner ( 2007 ) suggests that this ( scope-sp litting) r eading r esults fr om the co v ert mo v ement of the modal v erb can into a position betw een the negation and the univ ersal quantifier , as sho wn in ( 170 ). N ote the obligatory natur e of this modal mo v ement. The surface scope r eading (i.e., ‘not > 16. See H all ( 2015 ) f or a detailed ev aluation. 137 4.2. (N O N -)OCCURREN CE O F SEMANTI C EFFECTS WITH HEAD M O VEMENT : AN O N GO IN G D EB A TE ev ery > can ’) is a v ailable w hen the modal v erb can r econstructs. Thus, scope ambiguity is due to the optionality of modal r econstruction. (170 ) The assumed LF of ( 169 ), w her e can undergoes mo v ement to a higher position 17 [ N egP N ot [ can [ TP ev ery bo y [ T’ t can [ VP t ev ery bo y mak e the team ] ] ] ] F or this argument to go thr ough, it is crucial that the subject does not r econstruct back to the base, VP - internal position, as illustrated in ( 171 ); otherwise, ther e is no need to p osit modal mo v ement to deriv e the r elev ant scope r eading. T o justify this, Lechner assumes that str ong ( univ ersal) quantifier D P s c annot r econstruct belo w raising v erbs (i.e., belo w the modal v erbs ). (171) The LF of ( 169 ), w her e every boy r econstructs 18 [ N egP N ot [ [ TP <ev ery bo y> [ T’ can [ VP ev ery bo y mak e the team ] ] ] ] H o w ev er , it is pr ecisel y this assumption on str ong quantifier D P s that ar ouses contr o v ersies. H all ( 2015 ) and M c Closk ey ( 2016 ) specificall y argue against this assumption ( among other assumptions ). 19 Another r eported case of the scope eff ects of modal mo v ement concerns the interpr etation of modal v erbs and negation in English. I t is observ ed that a gr oup of deontic modals acr oss languages consistentl y r esist being interpr eted within the scope of negation, i.e., they ar e P ositiv e P olarity I tems, or P P I modals (I srael 1996 ; I atridou and Zeijlstra 2013 ; H omer 2015 ). Assuming that deontic modals ar e base-generated belo w negation, I atridou and Zeijlstra ( 2013 ) suggests that the □ > NEG r eading in the sentences in ( 172 ) is due to head mo v ement of the modal o v er negation. 20 The diff er ence in the surface w or d or der is due to the fact that modal mo v ement is o v ert in English, but co v ert in Gr eek and Spanish. As such, this modal mo v ement shifts the scope betw een deontic modals and negation. 21 17. N egativ e quantifiers ar e assumed to be licensed b y an abstract negation operator in the clause. F or simp licity , I indicate the negation scope b y putting not in a high N egP . I t is not meant to indica te negation mo v ement. 18. Angle brack ets < ... > indicate the surface position. 19. H o w ev er , Lechner ( 2017 ) argues that the v alidity of the argument need not hinge on the v alidity of the assumption on str ong quantifiers. H e also pr esents further evidence fr om comparativ es in support of modal mo v ement with scope eff ects. 20. The squar e symbol □ indicates the ( deontic ) necessity modals. 21. See also M atushansky ( 2006 ) f or r elev ant discussions. 138 4.2. (N O N -)OCCURREN CE O F SEMANTI C EFFECTS WITH HEAD M O VEMENT : AN O N GO IN G D EB A TE (172) (I atridou and Zeijlstra 2013 , p.530, 550 ) P P I modals must be interpr eted abo v e negation a. English,□ > NEG J o hn must / shoul d not lea v e. b. Gr eek,□ > NEG O the Y anis Y anis dhen neg pr epi must na N A figi. lea v e ‘ Y anis m ust not lea v e. ’ c. Spanish,□ > NEG J uan J uan no neg debe must ir . go ‘J uan must not go. ’ C iting the anal y sis pr oposed in H omer ( 2015 ), M c Closk ey ( 2016 ) argues that an “ alternativ e anal - y sis ... depends not on head mo v ement ... but rather on scope-enhancing co v ert mo v ement of the r ele- v ant modals. ” (p.9) and that “ appeal to head- raising is futile and they pr opose co v ert scope-expanding mo v ement, of the kind that H omer appeals to also f or English ” (p.9). In other w or ds, the mo v ement in v ol v ed in ( 172 ) is not head mo v ement per se , but some scope-expanding mo v ement ( e.g., Quantifier Raising). A ccor ding to M c Closk ey , this mo v ement does not constitute a case f or head mo v ement with scope eff ects. 22 4.2.3.4 M o v ement of aspectual v erbs The last case concerns the interpr etation of aspectual v erbs. S zabolcsi ( 2010 , 2011 ) r eports that in Shupamem (Bantu, S V O ), aspecctual v erbs can be optionall y be fr onted o v er a (quantificational) sub- ject and tak e scope in the deriv ed position. 23 The sentence in ( 173 a ) is the baseline examp le w her e yeshe ‘begin ’ is in the lo w position. Cruciall y , it raises o v er the subject in ( 173 b ) and scopes o v er the subject ‘ onl y M aria’ . Both sentences con v ey an unambiguous surface scope r eading. 22. I t seems that the argument her e r elies on a distinction betw een the notion of H ead M ovement as a distinct syntactic operation and the notion of head movement as a descriptiv e term f or all instances of mo v ement of a head. M y understanding of M c Closk ey’ s suggestion is that the case of modal mo v ement is not an instance of H ead M ovement , but, some other mo v ement operation such as Quantifier Raising. H o w ev er , the case of modal mo v ement still constitutes an instance of head movement with scope eff ects. 23. S zabolcsi ( 2010 , 2011 ) r eports similar scope eff ects in v erb/V2 mo v ement in Dutch. See discussions in § 4.7.3 . 139 4.2. (N O N -)OCCURREN CE O F SEMANTI C EFFECTS WITH HEAD M O VEMENT : AN O N GO IN G D EB A TE (173) Aspectual v erbs in Shupamem (S zabolcsi 2010 , p.38) a. ‘ onl y’ > ’begin ’ N dùù onl y M aria M aria ka past y eshe begin inget ha v e.inf ndàà good liP . r oles ‘Onl y M ary is such that she began to get good r oles’ b. ‘begin ’ > ‘ onl y’ A it -foc ka past y eshe begin ndùù onl y M aria M aria inget inf.ha v e ndàà good liP . r oles ‘I t b egan to be the case that onl y M ary is getting good r oles’ M c Closk ey ( 2016 ) suggests that if, f ollo wing S zabolcsi, aspectual v erbs ar e quantifiers o v er time v ari - ables, then it is expected to see the same scope eff ects with heads that in v ol v e quantification o v er w orld v ariables, e.g., modals, w hich, he suggests, has not been documented so far . 24 This r epr esents a concern on the generality of head mo v ement with scope eff ects. I t app lies to pr oposals on modal mo v ement w hich consistentl y ex clude a comparison with aspectual v erbs. 4.2.4 Interim summary T o sum up, I ha v e r eview ed cases of head mo v ement that do not seem to ha v e an y semantic eff ects, and they ar e tak en to be evidence that this is a general empirical pr operty of head mo v ement. H o w ev er , man y instances of head mo v ement bey ond the cor e inflectional sy stem ar e argued to impose semantic eff ects such as discourse eff ects or scope eff ects. These suggestions ar e not uncontr o v ersial, either because ther e ma y be alternativ e anal y ses to a head mo v ement appr oach, or because the pr oposed head mo v ement lacks generality . Against this back gr ound, I no w turn to the cor e empirical f oundations of this chapter , w her e aspectual v erbs and ( a subset set of ) modals in Cantonese, I argue, can undergo scope-shifting head mo v ement. 24. In fact, S zabolcsi ( 2011 , p.21) giv es one examp le of the modal counterpart in Shupamem. 140 4.3. THE D ISTRIB UTI O N O F ASP ECTU AL VERBS AND M O D AL VERBS 4.3 The distributi on of aspectual v erbs and modal v erbs This section pr esents basic data concerning the distribution of aspectual v erbs and modal v erbs in Cantonese. § 4.3.1 f ocuses on the a v ailability of a pr e-subject (high ) position f or these elements and corr esponding interpr etiv e eff ects. § 4.3.2 examines the types of v erbs that can appear in the high position. § 4.3.3 and § 4.3.4 discuss, r espectiv el y , tw o licensing conditions of the high position of as- pectual v erbs and modal v erbs. I t is r ev ealed that the pr esence of quantificational elements or f ocused elements is crucial in licensing the high position. I establish a distributional corr elation betw een quan- tificational/ f ocused elements and the v erbs in the high position, giv en in ( 174 ). (174) Generalization on the high position of aspectual v erbs and deontic/ futur e modals A high position of aspectual v erbs and deontic/ futur e modals is licensed iff the constituent that immediatel y f ollo w s this position (i) is quantificational or (ii) r eceiv e a f ocus interpr etation. 4.3.1 The ( r estri cted) high positi on The canonical position of aspectual v erbs and modal v erbs is a post -subject one. As alr eady seen in ( 152 ), the aspectual v erb hoici ‘begin ’ , f or examp le, can appear after the subject (i.e., the lo w position ), not b ef or e it (i.e., the high position ), r epeated belo w as ( 175 ). (175) = ( 152 ) The canonical lo w position of houci ‘begin ’ a. Aaming Aaming hoi ci begin haau -dou get -able hou good singzik. r esult ‘ Aaming begins to get good r esults. ’ b. * H oi ci begin Aaming Aaming haau get -able dou good hou r esult singzik. Int: ‘ Aaming begins to get good r esults. ’ 141 4.3. THE D ISTRIB UTI O N O F ASP ECTU AL VERBS AND M O D AL VERBS H o w ev er , the high position f or hoici ‘begin ’ is possible under certain cir cumstances. F or examp le, if the subject is mark ed with dak ‘ onl y’ , then hoici ‘begin ’ can appear in either the high or lo w position. (176) = ( 151 ) The lo w and high positions of houci ‘begin ’ a. ‘ onl y’ > ‘begin ’ / *‘begin ’ > ‘ onl y’ Dak onl y Aaming Aaming hoi ci begin haau -dou get -able hou good singzik. r esult ‘Onl y Aaming is such that he begins to get good r esults. ’ b. *‘ onl y’ > ‘begin ’ / ‘begin ’ > ‘ onl y’ H oi ci begin dak onl y Aaming Aaming haau -dou get -able hou good singzik. r esult ‘I t begins to be the case that onl y Aaming is getting go od r esults. ’ N o tabl y , the position of hoici ‘begin ’ indicates diff er ent scope r elations with dak ‘ onl y’ . In ( 176a ), hoici ‘begin ’ unambiguousl y tak es scope belo w ‘ onl y’ , w her eas in ( 176b ) it unambiguousl y scopes abo v e ‘ onl y’ . H er e, it is instructiv e to see ho w these scope r eadings ar e truth-conditionall y independent of each other . Consider the f ollo wing tw o scenarios in T able 4.2 , w hich concern the exam r esults in a class of thr ee student (building on the scenario first discussed in S zabolcsi ( 2010 , 2011 )). W ho is getting good r esults... Scenario 1 Scenario 2 T est 1 T est 2 T est 1 T est 2 Aaming: 40 Aaming: 100 Aaming: 100 Aaming: 100 B ill: 40 B ill: 40 B ill: 40 B ill: 40 Chris: 100 Chris: 100 Chris: 100 Chris: 40 Ü ( 176a ) onl y Aaming > begin Ü ( 176b ) begin > onl y Aaming T able 4.2: T w o scenarios of exam r esults in a class of thr ee In Scenario 1, among all students, Aaming is the onl y student w ho obtains an impr o v ed r esult in T est 2, w hile other students ar e doing as good/bad as bef or e. This scenario is true of ( 176a ), i.e., onl y Aaming is such that he begins to get good r esults. This is not true of ( 176b ). In Scenario 2, Aaming perf orms as good as bef or e in T est 2. H o w ev er , Chris, w ho w as doing gr eat in T est 1, perf orms not so w ell in 142 4.3. THE D ISTRIB UTI O N O F ASP ECTU AL VERBS AND M O D AL VERBS T est 2. This r enders Aaming b eing the onl y person w ho obtain good r esults in the class. This scenario is true of ( 176b ), i.e., it begins to be the case that onl y Aaming is getting good r esults. This is not true of ( 176a ). The sentences in ( 176 ) ar e thus not onl y unambiguous but also truth-conditionall y distinct. In other w or ds, the high position, if a v ailable, enables a v erb to tak e scope ther e o v er other struc - turall y lo w er elements. In all f ollo wing r elev ant cases, unless otherwise specified, I will onl y indicate the surface scope r eading in the English translation. 4.3.2 V erbs that can appear in the high positi on This subsection f ocuses on the types of v erbs that can appear in the high position. The cor e observ ation is that onl y raising pr edicates can occup y the high position. In all cases, I will contrast sentences with and without dak ‘ onl y’ . H o w ev er , it should be noted that dak ‘ onl y’ is not the onl y element that can license the high position. I postpone further discussions on the licensing conditions to § 4.3.3 and § 4.3.4 . 4.3.2.1 A spectual v erbs In addition to hoici ‘begin ’ , the same pattern is observ ed with gaizuk ‘ continue’ in ( 177 ). In ( 177 ) ( and all the subsequent examp les in this subsection ), the ( a ) sentences serv e as the baseline, w her e the lo w position is insensitiv e to the pr esence/ absence of dak ‘ onl y’ . Cruciall y , the (b ) sentences indicate that the high position is a v ailable if the subject is mark ed b y dak ‘ onl y’ . N ote that some examp les ar e giv en in embedded contexts to sho w that the high position is insensitiv e to r oot/ embedded en vir onments. (177) The aspectual v erb gaizuk ‘ continue’ and the high position a. N go I tenggong hear [(dak ) onl y H oenggong H ong.K ong gaizuk continue paai rank tau initial sap w ai ]. tenth ‘I hear d that ( onl y ) HK is such that she continues to rank among the top ten. ’ 143 4.3. THE D ISTRIB UTI O N O F ASP ECTU AL VERBS AND M O D AL VERBS b. N go I tenggon hear [ gaizuk continue * (dak ) onl y H oenggong H ong.K ong paai rank tau initial sap w ai ]. tenth ‘I hear d that it continues to be the case that ( onl y ) HK ranks among the top ten. ’ A dditionall y , a less discussed aspectual pr edicate, si-gwo ‘tried’ , sho w s the same pattern. In mor - p hological terms, si-gwo consists of the v erb si ‘try’ and the experiential suffix -gwo , but it has arguabl y undergone lexicalization and become a pr edicate (Shi, W ang, and Zhu 2002 ; W u 2020 ). M eaning - wise, it acquir es the meaning of ‘ha v e a certain experience’/‘ once’ , in addition to the original meaning of ‘try’ . 25 (178) The lexicalized aspectual pr edicate si-gwo ‘tried’ and the high position a. (Dak ) onl y ni -dou this-p lace si -gw o try -exp linzuk consecutiv el y lok fall sap- jat ten-da y jyu. rain ‘( Onl y ) t his p lace once rained f or ten consecutiv e da y s. ’ b. S i -gw o try -exp *(dak ) onl y ni -dou this-p lace linzuk consecutiv el y lok fall sap- jat ten-da y jyu. rain ‘I t w as once the case that onl y this p lace rained f or ten consecutiv e da y s. ’ A common pr operty shar ed b y hoici ‘begin ’ , gaizuk ‘ continue’ and si-gwo ‘tried’ is that all of them can be used as raising pr edicates, i.e., pr edicates that do not select an external argument. 26 This is in contrast with contr ol pr edicates such as soengsi ‘try’ and kyutding ‘ decide’ . The high position is disallo w ed, no matter w hether the subject is associated with dak ‘ onl y’ or not. 25. One of the signatur e pr operties of such aspectual usage is its compatibility with inanimate subjects (f or further discussions, see W u 2020 ). 26. F or discussions on the M andarin counterparts of ‘begin ’ and ‘ continue’ , see Y .-H. A. Li ( 1990 ). I assume the same app lies to Cantonese. 144 4.3. THE D ISTRIB UTI O N O F ASP ECTU AL VERBS AND M O D AL VERBS (179) Contr ol pr edicates cannot occup y the high position a. (dak ) onl y Aaming Aaming f soengsi / try kyutdingg decide tai r ead ni -bun this-cl syu. book ‘( Onl y ) Aaming tries to/ decides to r ead this book. ’ b. f * Soengsi / try * K yutdingg decide (dak ) onl y Aaming Aaming tai r ead ni -bun this-cl syu. book Int.: ‘( Onl y ) Aaming tries to/ decides to r ead this book. ’ 4.3.2.2 M odal v erbs F or the deontic modal v erb hoji ‘ma y’ and the futurity modal v erb wui ‘will’ in ( 180 ), the high position is a v ailable in the pr esence of dak ‘ onl y’ . N ote that the high position can be embedded in the comp lement clause of the pr eposition deoi ‘to ’ . (180 ) The deontic modal hoji ‘ma y’ and t he futur e modal wui ‘will’ a. N go I deoi to [ gamjat toda y (dak ) onl y Aaming Aaming f hoji / ma y / wui will g zou earl y fan ] sleep mou not.ha v e jigin. opinion ‘I ha v e no op inion on ( the claim that ) ( onl y ) Aaming ma y / will sleep earl y toda y . ’ b. N go I deoi to [ gamjat toda y f hoji / ma y / wui will g *(dak ) onl y Aaming Aaming zou earl y fan ] sleep mou not.ha v e jigin. opinion ‘I ha v e no opinion on ( the claim that ) it is allo w ed/it will be the case that ( onl y ) Aaming sleeps earl y toda y . ’ In contrast, modal v erbs r elating to ability , such as sik ‘be.able’ , and v olition, such as gaam ‘ dar e’ , fail to occup y the high position, no matter w hether the pr esence of dak ‘ onl y’ is pr esent or not, as sho wn in ( 181 ). 145 4.3. THE D ISTRIB UTI O N O F ASP ECTU AL VERBS AND M O D AL VERBS (181) M odal v erbs concerning ability and v olition fail to occup y the high position a. (Dak ) onl y Aaming Aaming f sik / be.able/ gaamg dar e tai r ead ni -bun this-cl syu. book ‘( Onl y ) Aaming is able/ dar e to r ead this book. ’ b. f * S ik / be.able * Gaamg dar e (dak ) onl y Aaming Aaming tai r ead ni -bun this-cl syu. book Int.: ‘( Onl y ) Aaming is able dar e to r ead this book. ’ N o te that the sp lit betw een deontic/ futur e modal v erbs and modal v erbs r elating to ability / v olition is not one betw een r oot and non- r oot modals. F ollo wing Lin and T ang ( 1995 ), B hatt ( 1998 ), and W urmbrand ( 1999 ), deontic/ futur e modals can be r egar ded as raising pr edicates, as opposed to modal v erbs r elating to ability / v olition ( w hich ar e r egar ded as contr ol pr edicates ). This sho w s that raising pr edicates, but not contr ol pr edicates, can occup y the high position. A comp lication arises if epistemic modals such as honang ‘possible’ and jinggoi ‘ should’ ar e tak en into consideration. W hile honang can occup y the pr e-subject position, it does not r equir e the pr esence of dak ‘ onl y’ . This is diff er ent fr om the pr evious cases of the high position f or aspectual v erbs and deontic/ futur e modals. (182) (Lin 2011 , p.51) The epistemic modal honang and the p r e-subject position a. Aaming Aaming honang be.possible zyu -gan cook -pr og faan. rice ‘ Aaming ma y be cooking. ’ b. H onang be.possible Aaming Aaming zyu -gan cook -pr og faan. rice ‘I t is possible that Aaming is cooking. ’ 146 4.3. THE D ISTRIB UTI O N O F ASP ECTU AL VERBS AND M O D AL VERBS The same can be said to the modal jinggoi ‘ should’ on its epistemic r eading. 27 (183) Epistemic jinggoi ‘ should’ and the pr e-subject position a. (Dak ) onl y Aaming Aaming camjat y ester da y jinggoi should lai -gw o come-exp ngo m y ukk ei. home ‘( Onl y ) Aa ming should ha v e come to m y home y ester da y . ’ b. J inggoi should (dak ) onl y Aaming Aaming camjat y ester da y lai -gw o come-exp ngo m y ukk ei. home ‘I t should be the case that ( onl y ) Aaming case to m y home y ester da y’ 4.3.2.3 Interim summary T o sum up, it is observ ed that the ( r estricted) high position is a v ailable f or raising pr edicates including aspectual v erbs and ( some ) modal v erbs, w hen the subject is associated with dak ‘ onl y’ . H o w ev er , epistemic modals can occup y the same position fr eel y , in an unr estricted w a y . N ote that I distinguish the r estricted high position fr om the unr estricted one ( with epistemic modals ), ev en though both of them ar e descriptiv el y a pr e-subject position. (184) The types of v erbs in the high position a. Onl y raising pr edicates can occur in the ( r estricted) high position. b. Epistemic modals can fr eel y occup y the pr e-subject position. 27. The modal v erb jinggoi ‘ should’ is ambiguous betw een an epistemic r eading and a deontic r eading. If it is interpr eted deonticall y , i t is pr edicted that it r equir es the pr esence of dak ‘ onl y’ to occur in the high position. F or examp le, a deontic r eading should be lacking in (i) due to the absence of dak . (i) J inggoi should Aaming Aaming lai. come ?Deontic r eading: ‘ Aaming should come. ’ O K Epistemic r eading: ‘I t is pr obable that Aaming comes. ’ Anticipating the discussions in § 4.3.4 , f ocused elements also license the high position. The deontic jinggoi ma y occup y the high positin b y virtue of the f ocus interpr etation of the subject Aaming . T o a v oid potential comp lications due to the pol y sem y of jinggoi , I f ocus on t he unambiguous deontic modal hoji ‘ma y’ , as discussed in ( 180 ). 147 4.3. THE D ISTRIB UTI O N O F ASP ECTU AL VERBS AND M O D AL VERBS 4.3.3 Quantifi cati onal el ements This subsection and the next examine the licensing conditions of the high position. The upshot is that the a v ailability of the high position depends on the natur e of the constituent that immediatel y f ollo w this position. In this subsection, I sho w that the high position is licensed if the r elev ant con- stituent is quantificational . This is consistent with the data in § 4.3.2 , w her e the pr esence/ absence of the quantificational element dak ‘ onl y’ is crucial to licensing the high position. In w hat f ollo w s, it is further sho wn that (i) the licensing elements can be quantificational elements other than dak ‘ onl y’; and (ii) the r elev ant constituent is not confined to subjects, but ma y tak e v arious f orms, ranging fr om topics to adv erbials and clauses . The general pattern is schematized belo w , w her e XP ranges o v er diff er ent constituents. (185) The high position is licensed b y the (immediatel y ) f ollo wing quantificational element a. O K A sp./M od. [ TP / CP XP [+quantificational] ... b. * A sp./M od. [ TP / CP XP [-quantificational] ... F irst, I start with diff er ent quantificational elements in the subject position. F or examp le, the high position of gaizuk ‘ continue’ is licensed if the subject is an ( existential) quantifier ( 186 a ), as opposed to a pr onoun ( 186 b ). R ecall that a pr oper name ( e.g., Aaming ) without dak ‘ onl y’ also fails to license the high position. (186) Gr oup denoting quantifiers v s. pr onouns in the subject position 28 29 a. Gaizuk continue f sub j jau ha v e jat -go one-cl jan person g haau -dou get -able hou good singzik. r esult ‘I t c ontinues to be the case that one person is getting good r esults. ’ 28. The underline in the examp les indicates the canonical position of the aspectual v erbs or modals in the high position. The same app lies to all subsequent examp les. 29. F or con v enience, I ha v e brack eted jau ‘ha v e’ and jat-go jan ‘ one person ’ to sho w its status as a subject. Indeed, it is mor e common to tr eat them as parts of an existential construction (H uang 1987 ; P aul 2021 ). This also app lies to the sentence in ( 187 a ). 148 4.3. THE D ISTRIB UTI O N O F ASP ECTU AL VERBS AND M O D AL VERBS b. * Gaizuk continue f sub j k eoi he g haau -dou get -able hou good singzik. r esult Int.: ‘I t continues to be the case that he is getting good r esults. ’ The sentences in ( 187 ) illustrate the same point with other quantificational elements. The high position is licensed b y negativ e quantifiers, counting quantifiers and wh -expr essions. (187) Diff er ent quantificational elements in the subject position a. H oji ma y f sub j mou not.ha v e jan person g lai come hoiwui. meeting ‘I t is allo w ed that no one comes to the meeting. ’ b. H oji ma y f sub j zuido at.most saam-go thr ee-cl jan person g lai come hoiwui. meeting ‘I t is allo w ed that at most thr ee peop le come to the meeting. ’ c. H oji ma y f sub j geido ho w .man y jan person g lai come hoiwui? meeting ‘H o w man y peop le ar e allo w ed to come to the meeting?’ Second, a similar contrast is observ ed in the topic position. Aspectual v erbs and modals ar e al - lo w ed to occup y the position higher than the topic if the topic is quantificational, but not if the topic is a ( non-quantificational) definite expr ession. 30 (188) U niv ersal quantifiers v s. definite NP s in the topic position a. H oi ci begin f to p cyunbou ev ery jan person g Aaming Aaming dou all hou v ery jansoeng. praise ‘I t b egins to be the case that Aaming praises ev ery one. ’ 30. The adv erb dou ‘ all’ is str ongl y pr ef err ed in the pr esence of univ ersal qua ntifiers. 149 4.3. THE D ISTRIB UTI O N O F ASP ECTU AL VERBS AND M O D AL VERBS b. * H oi ci begin f to p ni -go this-cl jan person g Aaming Aaming hou v ery jansoeng. praise Int.: ‘I t begins to be the case that Aaming praises this person. ’ Thir d, locativ e or frame-setting adv erbials can occur abo v e the subject. The high position of hoici ‘begin ’ (i.e. the position abo v e the locativ e/ frame-setting adv erbial) is allo w ed if the adv erbial is quan- tificational. (189) Quantificational v s. non-quantificational adv erbs a. H oi ci begin f ad v hai at mui -gaan ev ery -cl hokhaau school g Aaming Aaming dou all haau -dou get -able hou good singzik. r esult ‘I t b egins to be the case that at ev ery school, Aaming is getting good r esults. ’ b. ?? H oi ci begin f ad v hai at ngodei our hokhaau school g Aaming Aaming haau -dou get -able hou good singzik. r esult ‘I t b egins to be the case that at our school Aaming is getting good r esults. ’ If ther e is mor e than one quantificational element in the sentence, ther e ar e multip le possible positions f or hoici , and they deliv er diff er ent scope r eadings. (190 ) Quantificational adv erbs and subjects and multip le high positions a. f ad v H ai at mui -gaan ev ery -cl hokhaau school g hoi ci begin f sub j daaiboufan most jan person g dou all haau -dou get -able hou good singzik. r esult ‘ A t ev ery s chool, it begins to be the case that most peop le ar e getting good r esults. ’ ‘ ev ery’ > ‘begin ’ > ‘most’ 150 4.3. THE D ISTRIB UTI O N O F ASP ECTU AL VERBS AND M O D AL VERBS b. H oi ci begin f ad v hai at mui -gaan ev ery -cl hokhaau school g f sub j daaiboufan most jan person g dou all haau -dou get -able hou good singzik. r esult ‘I t begins to be the case that, at ev ery school, most peop le ar e getting good r esults. ’ ‘begin ’ > ‘ ev ery’ > ‘most’ A dditionall y , a similar contrast betw een quantificational and non-quantificational elements can be illustrated with subor dinate clauses. The sentences in ( 191 ) contain an if -clause and a whenever -clause. They can be r egar ded as quantificational elements since they quantify o v er possible w orlds. Cruciall y , they license the high position of hoici ‘begin ’ . T o facilitate compr ehension, I pr o vide a r elev ant context f or each sentence. (191) I f -clauses and whenever -clauses licensing the high position a. Context: the speaker is r eporting a r ecent mutational change on tr ees: decr ease in temper atur e leads to their leaves turning r ed. Pr eviously, decr ease in temper atur e did not necessarily lead to this r esult. H oi ci begin [ CP jyugw o if zyun become laang], cold syujip lea v es zau then wui will bin turn hung. r ed ‘I t begins to be the case that, if it becomes cold, the lea v es will turn r ed. ’ b. Context: the speaker is r eporting a r ecent climate change: it now becomes a certain f act that an appr oaching typhoon is associated with the appear ance o f big waves on the sea. Pr eviously, this association is only a matter o f chance. H oi ci begin [ CP faanhai w henev er daa- fung], appr oach-typ hoon hoimin sea dou all wui will jau ha v e daailong. big. w a v es ‘I t begins to be the case that, w henev er typ hoons appr oach, ther e will be big w a v es on the sea. ’ 151 4.3. THE D ISTRIB UTI O N O F ASP ECTU AL VERBS AND M O D AL VERBS Cruciall y , these subor dinate clauses ar e in contrast with with the non-quantificational although - clauses and because -clauses. N either of them license the high position of hoici ‘begin ’ . (192) because -clauses and although -clauses fail to license t he high position a. * H oi ci begin [ CP jaujyu because di cl linggin part loufaa ], aging bou cl dinnou computer w aaiw aaidei. out.of.or der Int.: ‘I t begins to be the case that, because the parts ar e aging, the computer is out of or der . ’ b. * H oi ci begin [ CP seoijin although mou not daa- fung], appr oach-typ hoon hoimin sea dou all jau ha v e daailong. big. w a v es Int.: ‘I t begins to be the case that, although no typ hoon is appr oaching, ther e ar e big w a v es on t he sea. ’ Based on these observ ations, the distribution of aspectual v erbs and deontic/ futur e modals in Cantonese can be stated as f ollo w s in ( 193 ). (193) Licensing conditions of the high position, part 1 A high position of aspectual v erbs and deontic/ futur e modals is licensed if the constituent that immediatel y f ollo w s this position is quantificational. 4.3.4 F ocused el ements Another licensing condition of the high position concerns f ocused elements. In additional to quan- tificational elements, a high position is also licensed b y a f ocused element that immediatel y f ollo w the high position. I substantiate this claim b y adopting diff er ent f ocus marking devices and sho w that all of them ma y license the high position. The general pattern sho w s the schema in ( 194 ). (194) The high position licensed b y the (immediatel y ) f ollo wing f ocused element a. O K A sp./M od. [ TP / CP XP [+f ocus ] ... b. * A sp./M od. [ TP / CP XP [- f ocus ] ... 152 4.3. THE D ISTRIB UTI O N O F ASP ECTU AL VERBS AND M O D AL VERBS N o te that part of the observ ations in this subsection w er e first discussed in Y ip and Lee ( 2020 ) f or M a ndarin. I pr o vide the Cantonese counterparts belo w , w hich pattern with the M andarin data. F irst of all, a f ocus r eading can be contributed b y the copula hai . The high position is licensed if hai is associated with the subject as in ( 195 a ). I t is not, ho w ev er , licensed b y object f ocus, as in ( 195 b ). (195) Copula f ocus and the high position a. H oji ma y hai co p Aaming Aaming heoi go H oenggong. H ong.K ong ‘I t ma y be the case that Aaming (but not others ) goes to H ong K ong. ’ b. * H oji ma y Aaming Aaming hai co p heoi go H oenggong . H ong.K ong Int.: ‘I t ma y be the case that Aaming goes to H ong K ong (but not other p laces ). ’ Second, a contrastiv e f ocus r eading can be f or ced b y the f ollo wing continuation. In ( 196 a ), a con- trastiv e f ocus r eading is f or ced on the subject, w her e Aaming is contrasted with Aaf an . H o w ev er , if the contrastiv e f ocus falls in elements in the VP , the high position is no longer licensed. The sentence in ( 196 b ) is degraded. (196) Contrastiv e/ corr ectiv e f ocus and the high position a. S i -gw o try -exp (hai) co p Aaming Aaming haau -dou get -able daijat, first, m- hai not -co p Aafan. Aafan ‘I t w as once the case that Aaming got first p lace, not Aafan. ’ b. ?? S i -gw o try -exp Aaming Aaming (hai) co p haau -dou get -able daijat, first, m- hai not -co p daiji. second Int.: ‘I t w as once the case that Aaming got the first p lace, not the second. ’ Also, other f ocusing devices such as the ‘ ev en ’ - f ocus constructions can also license the high po- sition, as long as the f ocused element occupies a position higher than the v erb. In ( 197 a ), the subject 153 4.3. THE D ISTRIB UTI O N O F ASP ECTU AL VERBS AND M O D AL VERBS r eceiv es f ocus r eading, but in ( 197 b ), a fr onted object is f ocused and, in ( 197 c ), the v erb zou ‘ do ’ is f ocused. All these f ocused elements license the high position. (197) ‘E v en ’ - f ocus constructions and the high position a. S i -gw o try -exp lin ev en Aaming Aaming dou also haau - m-dou get - not -able hou good singzik. r esult ‘I t w as once the case that ev en Aaming failed to get good r esults. ’ b. N go I gok dak think [ wui will lin ev en ni -bun syu this-cl book Aaaming Aaming dou also mou not tai ]. r ead ‘I t hink that it could be the case that Aaming didn ’t r ead ev en this book. ’ c. H oi ci begin lin ev en zou do Aaming Aaming dou also zou -cou -saai. do- wr ong -all ‘I t begins to be the case that Aaming ev en did it all wr ong. ’ One mor e examp le concerns wh -expr essions, w hose pr esence also licenses the high position. This is in line with the suggestion in R ochemont ( 1986 ) that wh -expr essions bear inher ent f ocus r eading. (198) W h -expr essions and the high position H oji ma y bin-go w ho haa- nin next - y ear heoi go H oenggong? H ong.K ong Lit.: ‘I t is allo w ed that w ho goes to H ong K ong next y ear?’ I t should be noted that other inf ormation structural notions, such as topics, do not license the high position. The sentence in ( 199 a ) in v ol v es a topicalized object and the one in ( 199 ) in v ol v es a base generated topic. N either of them licenses the high position of the aspectual v erb or the modal v erbs. 154 4.3. THE D ISTRIB UTI O N O F ASP ECTU AL VERBS AND M O D AL VERBS (199) T opics do not license the high position a. * H oi ci begin ni -bun this-boo k syu, Aaming Aaming fr om jau beginning tau r ead-up tai - hei. Int.: ‘I t begins to be that case that Aaming r eads this book fr om the beginning. ’ b. * H oji ma y seoigw o, fruit Aaming Aaming m-sik not -eat pinggw o. app le Int.: ‘I t ma y be the case that, as f or fruits, Aaming doesn ’t eat app les. ’ The observ ations in this subsection can be summarized in ( 200 ). (200 ) Licensing conditions of the high position, part 2 A high position of aspectual v erbs and deontic/ futur e modals is licensed if the con stituent that immediatel y f ollo w s this position r eceiv e a f ocus interpr etation. 4.3.5 Interim summary T aking stock, the empirical observ ations r eported in this subsection ar e r epeated and summarized belo w . In § 4.3.1 , I established that ther e is a high position in the clause that allo w s certain v erbs to tak e scope fr om ther e. In § 4.3.2 , I sho w that onl y raising pr edicates ma y occur in the ( r estricted) high position. Epistemic modals can also occup y the pr e-subject position, but this patterning is uncon- strained. (201) The types of v erbs in the high position a. Onl y raising pr edicates can occur in the ( r estricted) high position. b. Epistemic modals can fr eel y occup y the pr e-subject position. In § 4.3.3 and § 4.3.4 , tw o licensing conditions of the high position, w er e br ought to light, combined in ( 202 ) and schematicall y r epr esented in ( 203 ). 155 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT (202) Licensing conditions of the high position (part 1 + part 2) A high position of aspectual v erbs and deontic/ futur e modals is licensed iff the constituent that immediatel y f ollo w s this position (i) is quantificational or (ii) r eceiv es a f ocus interpr etation. (203) The high position licensed b y the (immediatel y ) f ollo wing quantificational/ f ocused element a. O K A sp./M od. [ TP / CP XP [+quantificational/+f ocus ] ... b. * A sp./M od. [ TP / CP XP [-quantificational/- f ocus ] ... 4.4 Pr oposal: scope-shifting head mo v ement T o captur e the empirical observ ations on aspectual v erbs and modal v erbs in § 4.3 , I pr opose that the high position of aspectual v erbs and deontic/ futur e modals is deriv ed via head mo v ement. I further pr opose that the app lication of this head mo v ement is constrained b y Scope Econom y (F o x 2000 ), w hich dictates that scope-shifting operations must ha v e a semantic eff ect. 4.4.1 T w o components of the pr oposal Bef or e I detail the pr oposal, I assume that aspectual v erbs and deontic/ futur e modals ar e r aising pr ed- icates (Y .-H. A. Li 1990 ; Lin and T ang 1995 ; B hatt 1998 ; W urmbrand 1999 ), and that the embedded subject has to mo v e to Spec TP f or Case (Y .-H. A. Li 1990 ), as sho wn in the clausal structur e in ( 204 ). (204) The basic structur e of sentences with aspectual v erbs and deontic/ futur e modals [ TP S ubj i [ Asp P / M odP A sp./M od. [ v P t i V ( Obj) ]] ] As raising pr edicates, the base position of these v erbs is v P -external. I will assume f or illustrativ e purposes that a n aspectual v erb heads an Aspect Phrase and a modal v erb a M odal Phrase. These v erbs ma y alternativ el y head a v erbal pr ojection abo v e the main v erb (i.e., another v P ). 156 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT 4.4.1.1 Ov ert scope-shifting head mo v ement I pr opose that aspectual v erbs and modal v erbs can undergo o v ert head mo v ement to the high position, and tak e scope fr om ther e. (205) The pr oposed o v ert head mo v ement [ TP A sp./M od. [ TP S ubj i [ v P t i V ( Obj) ]] scope-shifting head mo v ement R egar ding the imp lementation of this head mo v ement, I tentativ el y assume that (i) the mo ving head adjoins to the r oot structur e right abo v e the quantificational or the f ocused element, and that (ii) this head mo v ement is not driv en b y a syntactic f eatur e, but b y interpr etation/ scope considerations, in a w a y similar to other scope-shifting operations such as Quantifier Raising (M a y 1977 , 1985 ; F o x 2000 ; Bobaljik and W urmbrand 2012 ). H o w ev er , this is not the onl y w a y to imp lement the pr oposed head mo v ement. I r eturn to other possibilities in § 4.4.3 . I i llustrate the pr oposal with the examp les in ( 176 ), r epeated in ( 206 ). (206) The lo w and high positions of hoici ‘begin ’ a. (‘ onl y’ > ‘begin ’ / *‘begin ’ > onl y ) Dak onl y Aaming Aaming hoi ci begin haau -dou get -able hou good singzik. r esult ‘Onl y Aaming is such that he begins to get good r esults. ’ b. (*‘ onl y’ > ‘begin ’ / ‘begin ’ > ‘ onl y’) H oi ci begin dak onl y Aaming Aaming haau -dou get -able hou good singzik. r esult ‘I t begins to be the case that onl y Aaming is getting go od r esults. ’ U nder the curr ent pr oposal, the sentence in ( 206b ) is deriv ed fr om the sentence in ( 206a ), w her e hoici ‘begin ’ mo v es to the high position and tak es scope o v er dak ‘ onl y’ . The deriv ation is illustrated with English glosses in ( 207 ). 157 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT (207) Deriving ( 206b ) fr om ( 206a ) under a head mo v ement appr oach [ TP begin [ TP onl y Aaming i [ v P t i get -good- r esult ]]] scope enrichment scope-shifting head mo v ement I t should be noted that the pr oposed head mo v ement can mo v e o v er diff er ent elements, as long as the element is quantificational or f ocused. The flexibility of the landing site can be schematicall y illustrated in ( 208 ). N ote that, f or illustrativ e purposes, the notation [+Q/+F] is added to indicate the quantificational or f ocus natur e of the r elev ant constituent. I am not committed to the pr esence of an y [+Q/+F] f eatur e in these elements. (208) A schematic r epr esentation of the pr oposed head mo v ement CP CP S ub. Cl. [+Q/+F] T opic/F ocus P T opic/F ocus T opic [+Q/+F] TP TP A dv erb [+Q/+F] TP TP S ubject [+Q/+F] Asp P /M odP A sp./M od. v P A potential concern of the pr oposal is that the pr oposed head mo v ement as illustrated in ( 208 ) appears to violate the H ead M o v ement Constraint (HM C, T ra vis 1984 ), since the mo ving head, as pr oposed, mo v es in one f ell sw oop and potentiall y cr osses other heads such as the T / C head or the T opic/F ocus head. 31 I suggest that violation to HM C does not pr eclude a head mo v ement anal y sis, since the status of HM C as a general syntactic constraint is not uncontr o v ersial. M an y instances of 31. F or a clear violation of HM C, see ( 232 ). 158 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT head mo v ement ha v e been r eported in the literatur e to sho w a clear violation to HM C, including (but not limited to ) Long H ead M o v ement (Lema and Riv er o 1990 ; Riv er o 1991 ; Borsley , Riv er o, and S tep hens 1996 , i.a.), and pr edicate clefts/ v erbal r eported in v arious languages (V icente 2007 ; H ein 2018 ; H arizano v 2019 , i.a.) fr onting. I suggest that the pr oposed head mo v ement falls into the famil y of head mo v ement that does not obey HM C. 32 A r emaining question is ho w this head mo v ement is constrained in a w a y such it onl y allo w s certain v erbs to occup y the high position under certain cir cumstances. I discuss this issue in the next subsection. 4.4.1.2 Scope Econom y T o account f or the constraints on the pr oposed head mo v ement, I suggest that onl y quantificational heads , or heads that ar e generalized quantifiers (i.e., functions fr om pr operties to truth v alues, cf. Bar - wise and Cooper ( 1981 )), can be targeted f or the pr oposed head mo v ement. S ubstantiall y , I pr opose a parallel quantificational anal y sis of aspectual v erbs and modals. On one hand, aspectual v erbs ar e generalized quantifiers o v er times ( of semantic type <,t>, i.e., functions fr om sets of time interv als to truth v alues ) ( cf. S zabolcsi 2010 , 2011 ). 33 On the other hand, deontic/ futur e modals ar e general - ized quantifiers o v er worlds ( of semantic type <,t>, i.e., fr om sets of w orlds to truth v alues ) ( cf. M atushansky 2006 ; v on F intel and H eim 2011 ; I atridou and Zeijlstra 2013 ). Aspectual v erbs Deontic/F utur e modal v erbs Cor e meaning generalized quantifiers o v er times generalized quantifiers o v er w orlds Semantic type <,t> <,t> T able 4.3: A parallel quantificational anal y sis of aspectual v erbs and modal v erbs Importantl y , I suggest that their mo v ement is subject to a constraint on ( scope-shifting) mo v ement 32. This is not to sa y that HM C should be abandoned, but that HM C appears to be too str ong as a general constraint on all instances of head mo v ement. R ef ormulation of HM C is much desir ed, but w ould go bey ond the scope of this chapter . See R oberts ( 2001 ), M atushansky ( 2006 ), and H arizano v and Gribano v a ( 2019 ), and Chapter 3 f or discussions. 33. This is not meant to unify diff er ent usages of aspectual v erbs – they can also be used as transitiv e pr edicates, f or examp le. The claim her e is that at least the usage of aspectual v erbs under discussion can r eceiv e a quantificational anal y sis. 159 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT of generalized quantifiers, Scope Econom y (F o x 2000 ). 34 F o x’ s original discussion of Scope Econom y f ocuses on nominal/ p hrasal quantifiers, but if Scope Econom y is a constraint on scope-shifting oper - ations in general, ther e is no a priori r eason to rule out its app lication to scope-shifting operations of v erbal/ non-p hrasal quantifiers. (209) Scope Econom y (F o x 2000 , p.23, modified) Scope-shifting operations must ha v e a semantic eff ect. This constraint is crucial in exp laining the generalization in ( 202 ). This is because, in the absence of a quantificational or a f ocused element (i.e., the unacceptable cases ), app l ying the pr oposed mo v ement w ould not shif t/ aff ect the r elativ e (quantificational or f ocus ) scope r elations. Bef or e I detail ho w the r elativ e scope is shifted/ aff ected in § 4.4.2 , the idea can be illustrated in ( 210 ). (210 ) The high position licensed b y the (immediatel y ) f ollo wing quantificational/ f ocused element a. O K A sp./M od. [ TP / CP XP [+quantificational/+f ocus ] ... ... M o v ement wit h semantic eff ects b. * A sp./M od. [ TP / CP XP [-quantificational/- f ocus ] ... ... M o v ement wit hout semantic eff ects T o see an examp le, in the sentences in ( 175 ), r epeated in ( 211 ), w her e the subject is non-quantificational (because it is a pr oper name ). In such case, the mo v ement of hoici ‘begin ’ fails to shift scope r elations, and thus violates Scope Econom y . The diagram in ( 212 ) illustrates this idea with English glosses. 34. W hile the original v ersion of Scope Econom y is a constraint on co v ert operations, T akahashi ( 2002 ) and M atyiku ( 2017 ) r eport cases w her e o v ert head mo v ement is constrained b y Scope Econom y , suggesting that Scope Econom y is not a constraint specific to co v ert syntax. See also Bobaljik and W urmbrand ( 2012 ) f or a similar suggestion. 160 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT (211) = ( 175 ) The canonical lo w position of houci ‘begin ’ a. Aaming Aaming hoi ci begin haau -dou get -able hou good singzik. r esult ‘ Aaming begins to get good r esults. ’ b. * H oi ci begin Aaming Aaming haau -dou get -able hou good singzik. r esult Int: ‘ Aaming begins to get good r esults. ’ (212) An attempted deriv ation of ( 175b ) fr om ( 175a ) under a head mo v ement appr oach [ TP begin [ TP Aaming i [ v P t i get -good- r esult ]]] no scope -shifting As such, a head mo v ement anal y sis, coup led with Scope Econom y , deriv es the basic paradigm discussed in ( 151 )/( 176 ) and ( 152 )/( 175 ). Some qualifications ar e in or der . I assume with F o x ( 2000 ) that Scope Econom y can be imp le- mented as a syntactic, deriv ational constraint on mo v ement ( cf. Bobaljik and W urmbrand 2012 , f or a similar stance ). In other w or ds, it is a condition that app lies in a local manner during a syntactic deriv ation. In eff ect, at each step of a deriv ation, it determines w hether a quantificational element can undergo mo v ement, b y checking w hether the mo v ement w ould potentiall y ha v e a semantic ef - f ect. F or examp le, w hen the non-quantificational subject Aaming is merged in Spec TP in ( 212 ), Scope Econom y determines that the mo v ement of ‘begin ’ w ould not ha v e a semantic eff ect, hence blocking its mo v ement. In contrast, w hen the quantificational subject dak Aaming ‘ onl y Aaming’ is merged in Spec TP in ( 207 ), Scope Econom y sanctions the mo v ement of ‘begin ’ f or its potential scope eff ects (it does not f or ce the mo v ement, ho w ev er ). 35 One immediate question r elating to the mechanics of Scope Econom y is ho w such determination is possible in syntax, pr o vided that semantic/ scope eff ects ar e commonl y assumed to be pr operties 35. I t should be noted that Scope Econom y ma y be suggested to be a r epr esentational constraint, w hich, f or examp le, app lies at each Spell -Out. 161 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT r ead off in the Logical F orm (LF). I again assume with F o x ( 2000 , p.66-74) in that a subset of f ormal logical pr operties ar e accessible to syntax, and Scope Econom y is calculated based on these pr operties. In princip le, it might be suggested that these f ormal logical pr operties ar e syntacticall y r epr esented, e.g., as a ( null) syntactic f eatur e on certain constituents. 4.4.2 Deri ving the pr operti es of mo v ement of quantifi cati onal heads In this subsection, I illustrate ho w the semantic eff ects of the pr oposed head mo v ement ar e achiev ed, w hich ar e crucial in licensing the pr oposed head mo v ement. I first illustrate in § 4.4.2.1 the quantifi- cational scope eff ects with aspectual v erbs, with special f ocus on compositionality . Then I illustrate in § 4.4.2.2 the f ocus scope eff ects with modal v erbs, emp hasizing the subtle but detectable eff ects on the calculation of f ocus sets. Lastl y , in § 4.4.2.3 , I sho w ho w the curr ent pr oposal deriv es the r estriction on v erbs based on considerations of semantic types. 4.4.2.1 Deri ving the quantifi cati onal scop e eff ects Bef or e I pr oceed, I assume basic compositional rules fr om H eim and Kratzer ( 1998 ) and a framew ork w her e time and w orld v ariables ar e exp licitl y intr oduced in the syntax (b y the Aspect head and the M o dal head, r espectiv el y ), suggested in Beck and S techo w ( 2015 ). F ollo wing S zabolcsi ( 2010 , 2011 ), the lexical semantics of hoici ‘begin ’ can be giv en in ( 213 ). I t indicates that ther e exist tw o time interv als t’ and t” such that t’ pr ecedes a contextuall y determined time v ariable, pr esumabl y the speech time, w hich pr ecedes t” . A pr oposition P is false at ( the earlier ) time t’ but it is true at ( the later ) time t” . (213) The lexical semantics of hoici ‘begin ’ (largel y based on S zabolcsi 2010 , 2011 ) J hoiciK = P .9 t’9 t” [ t’ < t* t”^ P( t’) = 0^ P( t”) = 1] ( w her e t* is a time v ariable w hose v alue is contextuall y determined) W ith all these ingr edients, the quantificational scope eff ects of mo v ement of hoici ‘begin ’ in sen- 162 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT tences lik e ( 176b ), r epeated in ( 214 ), can be illustrated in ( 215 ). After undergoing the pr oposed head mo v ement, hoici ‘begin ’ tak es scope in the high position o v er dak ‘ onl y’ in the subject position. N ote that subject mo v ement fr om the specifier of v P to the specifier of TP is due to Case r easons ( cf. Y .-H. A. Li 1990 ). F or simp licity , I assume that dak Aaming ‘ onl y Aaming’ f orms one constituent as a D P . (214) H oici ‘begin ’ in the high position =( 176b ) H oi ci begin dak onl y Aaming Aaming haau -dou get -able hou good singzik. r esult ‘I t begins to be the case that onl y Aaming is getting good r esults. ’ (215) A compositional anal y sis and the semantics of ( 176b ) a. TP t Asp begin <,t> T’ 2 T’ t D P [ onl y Aaming] T’ <e,t> 1 Asp P t Asp t 2 v P i v P t D P t 1 get -good- r esult b. J ( 215a )K =9 t’9 t” [ t’ < t* t”^ [ onl y Aaming x. get -good- r esult(x)( t’) = 0 ] ^ [ onl y Aaming x. get -good- r esult(x)( t”) = 1 ] ] As pr oposed her e, aspectual v erbs ( and deontic/ futur e modals ) ar e ( v erbal) generalized quantifiers. F ollo wing the anal y sis on nominal generalized quantifiers (H eim and Kratzer 1998 ), I s uggest that their mo v ement lea v es a trace of a lo w er type (i.e., a time/ w orld v ariable ) in a similar w a y that p hrasal 163 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT quantifiers ( of type <<e,t>,t>) lea v e behind an individual type trace in instances of mo v ement. W hen hoici mo v es, cruciall y , it lea v es behind a trace (i.e., t 2 in ( 215 a )), w hich is a time v ariable of type i. 36 H oici lands at a position right abo v e the subject and tak es scope ther e, deliv ering the meaning in ( 215b ). 37 S imilar can be said about deontic/ futur e modals, with the onl y diff er ence being the semantic types. The w orld v ariables w ould r ep lace the time v ariables. In the absence of a quantificational element, the mo v ement w ould not be licensed as it w ould not impose an y scope eff ects. As a r emark, S zabolcsi ( 2011 ) pr oposes an alternativ e semantic type f or aspectual v erbs, w her e they ar e of modifier type <,>. This is a possible option, but it must also come with additional assumptions on the mo v ement trace. U nder the curr ent appr oach, the trace cannot be a lo w type trace ( a time v ariable ), as it w ould lead to a potential type mismatch issue w hen the modifier type hoici ‘begin ’ tak es scope in the deriv ed position: it r eturns a function instead of a truth v alue. Diff er - ent solutions ha v e been suggested in the literatur e. F or examp le, the mo v ement ma y be an instance of trace- less mo v ement or the trace might be deleted at LF (Cable 2010 ; S tepano v 2012 ; M atyiku 2017 ). Alternativ el y , emp lo ying flexible types, aspectual v erbs ma y be type - lifted in case of mo v e- ment (M atyiku 2017 ) or they ma y undergo function composition with a type- lifted tense operator (S zabolcsi 2011 ). T o the extent that modals ar e mor e commonl y identified as generalized quantifiers than as modifiers, I adopt a unif orm anal y sis of both aspectual v erbs and modals. But the pr ecise choice betw een generalized quantifier types and modifier types does not bear on the pr oposal. In princip le, the pr oposed head mo v ement can target heads that ar e of either type ( see further discus- sions in § 4.7.2 ). 36. If hoici does not mo v e, it tak es scope at the position abo v e v P but belo w the subject in Spec TP . 37. One concern on meaning in ( 215b ) is that the first conjunct in the scope of ‘begin ’ pr esupposes that Aaming got good r esults as w ell at t’ , since dak ‘ onl y’ pr esupposes its pr ejacent to be true ( similar to English only ). This is ho w ev er not the intended r eading of ( 214 ). I ackno w ledge that this is a non-trivial issue f or an accurate f ormulation of the lexical semantics of hoici ‘begin ’ , but I ha v e to lea v e this issue to futur e r esear ch. 164 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT 4.4.2.2 Deri ving the f ocus scope eff ects I no w turn to the scope eff ects w hen modal v erbs ( and aspectual v erbs ) mo v e o v er a f ocused element. 38 . I t should be str essed that the pr oposed head mo v ement does not aff ect the f ocus scope, but the cal- culation o f the f ocus set , or the f ocus alternatives . 39 F ocus scope is determined independentl y b y the marking of the copula hai or lin ‘ ev en ’ , f or examp le. The pr oposed head mo v ement does not impose an eff ect on w hat is mark ed as f ocus. Instead, it aff ects the size of the f ocus set, w hose calculation r ests on the structural position of the modal v erbs and the f ocus operator . T o illustrate this idea, tw o assumptions ar e needed. I assume an alternativ e semantic framew ork on f ocus interpr etation (R ooth 1985 , 1992 ). Alternativ e semantics k eep track of both the or dinary semantic values and the f ocus semantic values ( or , equiv alentl y , the alternativ e set ) during the syntactic deriv ation. The f ocus semantic v alues consist of the set of all or dinary semantic v alues obtained b y r ep lacing alternativ es f or an y f ocus- mark ed sub-parts. T ak e ( 216 ) as an examp le, w hich is the baseline sentence f or ( 195 a ), i.e., the modal is in the base position. The sentence ma y r eceiv e a subject f ocus r eading or a sentential f ocus r eading. (216) Copula f ocus marking on the subject baseline of ( 195 a ) H ai co p Aaming Aaming hoji ma y heoi go H oenggong. H ong.K ong ‘I t ma y be the case that Aaming (but not others ) goes to H ong K ong. ’ The or dinary semantic v alue is the pr oposition in ( 216 ). On a subject f ocus r eading, the alterna- tiv e semantic v alue is the set of pr opositions obtained b y r ep lacing the f ocus- mark ed Aaming with other entities, such as Aaf an , J ohn , M ary , etc. S imilarl y , on a sentential f ocus r eading, the alternativ e semantic v alue is obtained b y r ep lacing the w hole pr oposition with other pr opositions. 38. The discussions in this subsection is based on Y ip and Lee ( 2020 ) 39. Y .-y . H su ( 2019 ) suggests that the mo v ement of modal v erbs to the high position giv es rise to a pr opositional f ocus r eading, but this is not necessaril y the case. F or examp le, the sentence in ( 195 a ) r etains a subject f ocus r eading. 165 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT (217) The or dinary and alternativ e semantic v alues of ( 216 ) 40 a. J ( 216 )K O = Aaming ma y go to H ong K ong. b. subject f ocus J ( 216 )K alt = 8 > > > > > > > < > > > > > > > : Aafan ma y go to H ong K ong. J ohn ma y go to H ong K ong. M ary ma y go to H ong K ong. ... 9 > > > > > > > = > > > > > > > ; c. sentential f ocus J ( 216 )K alt = 8 > > > > > > > < > > > > > > > : Aafan should come to US. J ohn must sta y in J apan. M ary w ants to lea v e T aiw an. ... 9 > > > > > > > = > > > > > > > ; The second assumption concerns the r ole of f ocus operators/ f ocus particles. F ollo wing Beck ( 2006 ), I suggest that a f ocus operator can look at the f ocus semantic v alues and “r eset” the f ocus semantic v alues of its comp lement b y r ep lacing the f ocus semantic v alues with the or dinary semantic v alues ( cf. R ooth ’ s operator; s ee also discussions in K otek ( 2016 , 2019 ) and Erlewine ( 2020a )). W ith these ingr edients, the scope eff ects of the mo v ement of the modal v erb in ( 216 ) can be il - lustrated in ( 218 ). F or simp licity , I f ocus on the subject f ocus r eading, but the same app lies to the sentential f ocus r eading. (218) F ocus scope eff ects of the pr oposed head mo v ement a. = ( 195 a ) H oji ma y hai co p Aaming Aaming heoi go H oenggong. H ong.K ong ‘I t ma y be the case that Aaming (but not others ) goes to H ong K ong. ’ b. J ( 218a )K O = Aaming ma y go to H ong K ong. 40. JK O is the interpr etation function f or an or dinary semantic v alue, w her eas JK alt is f or a f ocus semantic v alue. 166 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT c. subject f ocus J ( 218a )K alt = I t is allo w ed that 8 > > > > > > > < > > > > > > > : Aafan goes to H ong K ong. J ohn goes to H ong K ong. M ary goes t o H ong K ong. ... 9 > > > > > > > = > > > > > > > ; Cruciall y , the size of the alternativ e set in ( 218 c ) is diff er ent fr om that in ( 216 b ). The modal v erb hoji ‘ma y’ escapes the scope of the f ocus operator hai b y mo ving to the high position. As a r esult, the structural r elation betw een the modal v erbs and the f ocus operators changes, and so does as the size of the f ocus set. A straightf orw ar d pr ediction along this line of r easoning is that w hat can serv e as a f elicitous continuation diff ers in cases with and without mo v ement. This is borne out in the contrast r ev ealed in ( 219 ). In ( 219 a ), the modal v erb mo v es to the high position, and the continuation cannot f elicitousl y contain the same modal v erb hoji ‘ma y’ , as it is no longer within the scope of the f ocus operator . This is in contrast with the sentence in ( 219 b ), w her e the modal v erb does not undergo mo v ement, and the same continuation becomes f elicitous. (219) Diff er ences in f elicitous continuation with and without mo v ement of the modal v erb a. # H oji ma y hai co p Aaming Aaming heoi go H oenggong, H ong.K ong m- hai not -co p Aafan Aafan hoji ma y heoi. go Int.: ‘I t ma y be the case that Aaming goes to H ong K ong, but not that Aafan ma y go ( to H ong K ong). ’ b. H ai co p Aaming Aaming hoji ma y heoi go H oenggong, H ong.K ong m- hai not -co p Aafan Aafan hoji ma y heoi. go ‘ Aaming ma y go to H ong K ong, but not Aafan ma y go ( to H ong K ong). ’ I t is ob vious that hai is not the onl y f ocus operator in Cantonese. Cases with lin ‘ ev en ’ sho w the same pattern, as discussed in ( 197 ). F or cases without o v ert f ocus operators such as ( 196 ), I assume that ther e is a null f ocus operator . S ince the pr esence of f ocus operators is crucial in licensing the pr oposed 167 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT head mo v ement, it f ollo w s that topic structur es discussed in ( 199 ) do no t license the mo v ement. One r emaining issue is that, if the high position hinges on w hether the mo v ement alters scope r e- lations with the f ocus operator , then the unacceptability of ( 195 b ) might at first glance be surprising. This is because if the modal v erb is base generated belo w the f ocus operator (= ( 220 )) and undergoes mo v ement, it cr osses the f ocus operator . H o w ev er , ( 195 b ) is unacceptable. I suggest that the mo v e- ment is disallo w ed not because it violates Scope Econom y , but because it violates another locality constraint on mo v ement, Shortest M o v e. I postpone the discussion of these cases to § 4.5.2 . (220 ) A puzzle on the unacceptability of ( 195 b ) a. =( 195 b ) * H oji ma y Aaming Aaming hai co p heoi go H oenggong . H ong.K ong Int.: ‘I t ma y be the case that Aaming goes to H ong K ong (but not other p laces ). ’ b. baseline of ( 195 b ) Aaming Aaming hai co p hoji ma y heoi go H oenggong . H ong.K ong ‘ Aaming ma y go to H ong K ong (but not other p laces ). ’ 4.4.2.3 Deri ving the r estri cti on on v erbs R ecall the r estriction on the types of v erbs that can appear in the high position, r epeated belo w in ( 221 ). (221) The types of v erbs in the high position a. Onl y raising pr edicates can occur in the ( r estricted) high position. b. Epistemic modals can fr eel y occup y the pr e-subject position. I t is observ ed that the high position is not a v ailable to contr ol v erbs and ability / v olition modal v erbs. In other w or ds, these v erbs fail to undergo the same head mo v ement pr oposed f or aspectual v erbs and deontic/ futur e modals. I suggest that it is the semantic types of these v erbs that diff er entiate the 168 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT tw o gr oups. U nder the curr ent pr oposal, aspectual v erbs and deontic/ futur e modal v erbs ar e generalized quan- tifiers and ar e of type << ,t>,t>. This semantic type is possible f or raising pr edicates, as they do not tak e an y external argument but one clausal argument, i.e., they ar e functions fr om sets of pr operties to truth v alues. In contrast, contr ol v erbs and ability / v olition modals tak e tw o arguments, namel y , a clausal argument and an external argument. As such, they ar e functions fr om sets of pr opositions to functions fr om entity to truth v alue, f or examp le, of type <,<e,t>>. I suggest that the r eason w h y they cannot undergo the pr oposed head mo v ement is that ther e is no possible type of their trace that w ould r ender successful composition. F or examp le, if their corr esponding t race is of individual type ( e.g., a time v ariable or a w orld v ariable ), a type mismatch is una v oidable. Other higher type traces (if possible ) could not a v oid the type mismatch either . In or der to guarantee successful composition, the onl y possible type of the trace w ould be the same type as the mo ving head, i.e., <,<e,t>>. If this is the case, the mo ving head, ho w ev er , w ould fail to tak e scope in its deriv ed position; it is ( semanticall y ) r econstructed back to its base position. I ts mo v ement w ould then fail to shift scope and violate Scope Econom y . 41 Ther ef or e, contr ol v erbs and ability / v olition modal v erbs fail to undergo the pr oposed head mo v ement because they ha v e a seman- tic type substantiall y diff er ent fr om aspectual v erbs and deontic/ futur e modals, and their mo v ement w ould either lead to compositional conflicts or a violation of Scope Econom y . Inter estingl y , this exp lanation points to a pr edication on possible mo v ement of contr ol v erbs and ability / v olition modals. If the mo v ement of these heads is trigger ed not b y interpr etation/ scope con- siderations, but b y some syntactic f eatur e, their mo v ement should be allo w ed. This is because their trace can be of the same type, i.e., the mo v ement can r econstruct without violating Scope Econom y . This pr ediction is borne out. In Cantonese, it is argued that a v erb can be fr onted ( with doubling) to 41. U nlik e modifier types, it appears to be technicall y impossible to ensur e a wide scope r eading of contr ol v erbs and ability / v olition modal v erbs and successful composition at the same time. F or examp le, type- lifting the mo ving heads w ould lead to a higher type <<,<e,t>>, <,<e,t>>> in the high position, but then it w ould lack the r elev ant argu - ments f or further composition. Assuming trace- less mo v ement or LF deletion of traces w ould not help either since the composition w ould crash at the base position. 169 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT deliv er a topic r eading ( 222a ) or it can mo v e on its o wn to the end of the sentence to escape a f ocus interpr etation ( 222b ) (Cheng and V icente 2013 ; Lee 2017 ; L ai 2019 ). 42 (222) a. T opic constructions of v erbs contr ol v erbs Soengsi , try Aaming Aaming hai foc soeng w ant soengsi try tai r ead ni -bun this-cl syu. book ‘ As f o r t rying, Aaming w ants to try to r ead this book. ’ b. Right dislocation of v erbs ability modals Aaming Aaming tai r ead ni -bun this-cl syu book gaa3 sfp sik . be.able ‘ Aaming is able to r ead this book. ’ These observ ations suggest that it is not that contr ol v erbs and ability / v olition modal v erbs can nev er mo v e. They can mo v e, if r econstruction does not lead to an y violation of the grammar . H o w ev er , they cannot undergo the pr oposed mo v ement because r econstructing the mo v ement w ould violate Scope Econom y . T urning to the unr estricted pr e-subject position of epistemic modal v erbs, r ecall that they can fr eel y appear bef or e or after the subject, r egar dless of w hether the subject is quantificational or f o- cused. A t first glance, the distribution of epistemic modals appears to speak against the curr ent pr o- posal as the pr e-subject position is not r egulated b y Scope Econom y , i.e., epistemic modals appear to be immune fr om Scope Econom y . I suggest that this is because epistemic modals ha v e a diff er ent syntax compar ed to aspectual v erbs and deontic/ futur e modals, and that the pr e-subject position is not deriv ed via mo v ement. F ollo wing a r ecent pr oposal b y Lin ( 2011 , 2012 ), epistemic modals tak e a finite TP comp lement, as opposed to deontic/ futur e modals, w hich tak e a non- finite TP comp lement (i.e., t he latte r ar e raising pr edicates ). U nder this view , epistemic modals occurring in the pr e-subject position ha v e the base structur e in ( 223 a ). 42. F or extensiv e disccussions on these structur es, see Chapter 3. 170 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT (223) The diff er ence betw een epistemic and deontic modals under the pr oposal in Lin ( 2011 , 2012 ) a. ... [ M odP epistemic modals [ finite TP S ubj. V Obj. ] ] b. ... [ M odP deontic/ futur e modals [ non- finite TP S ubj. V Obj. ] ] Assuming that Chinese has Case and finite clauses can assign nominativ e Case in Chinese (Y .-H. A. Li 1990 ), the (post - modal) subject in ( 223 a ) is in a Case position, and it can sta y ther e without further mo v ement. This is diff er ent fr om deontic modals ( or other raising pr edicates ), w her e the subject is in a non- finite clause in ( 223 b ). I t needs to mo v e f or Case (i.e., the sentence is a raising structur e ). F or cases w her e the epistemic modals f ollo w the subject, this can be deriv ed via an optional mo v ement of the subject o v er the epistemic modal,e.g., trigger ed b y an EP P f eatur e (Lin 2011 ), or a topic f eatur e (Chou 2013 ; T sai 2015 ). In other w or ds, epistemic modals ha v e a non- raising structur es, and this exp lains w h y their pr e-subject position is less r estricted (because they ar e base-generated ther e ), as opposed to other raising pr edicates. This exp lanation of the less r estricted epistemic modals is corr oborated b y the distribution of sen- tential negation, w hich sho w s a similar pattern to epistemic modals. The high position of the sentential negation m-hai ‘not’ is unr estricted. (224) Sentential negation and the pr e-subject position a. (Dak ) onl y Aaming Aaming m - hai not -co p haau -dou get -able hou good singzik. r esults ‘( Onl y ) Aa ming didn ’t get good r esults. ’ b. M - hai not -co p (dak ) onl y Aaming Aaming haau -dou get -able hou good singzik. r esult ‘I t is not the case that ( onl y ) Aaming gets good r esults. ’ The facts f or both epistemic modals and sentential negation f ollo w if both of them can tak e a finite comp lement clause, w her e the subject can sta y in-situ. If these anal y ses ar e on the right track, their 171 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT unr estricted distribution is irr elev ant to Scope Econom y , since ther e is no mo v ement of epistemic modals ( or sentential negation ) in the first p lace. 4.4.3 R emar ks on the landing site and the trigger In the pr oposal, I ha v e assumed that the pr oposed head mo v ement is achiev ed b y adjunction to the r oot structur e and that the mo v ement is trigger ed b y interpr etation/ scope considerations. I t should be ackno w ledged that this is not the onl y w a y to imp lement the pr oposed head mo v ement. I discuss a number of possibilities, w hich r el y on diff er ent assumptions. H o w ev er , the ultimate choice among these options does not aff ect the central claim that head mo v ement can induce scope eff ects. 4.4.3.1 The landing site W ithin the rich literatur e on head mo v ement, almost all logicall y possible landing sites ha v e been pr o- posed. ( 225 ) illustrates ( with hoici ‘begin ’) f our possible landing site, namel y , ( a ) a head-adjoined posi - tion, (b ) a specifier position, ( c ) a deriv ed head position and, ( d) a p hrase-adjoined position, illustrated with hoici as f ollo w s, r espectiv el y . ( 225 a ) is achiev ed via head-to- head adjunction (Bak er 1988 , et seq. ) and ( 225 b ) is achiev ed via head-to-specifier mo v ement (M atushansky 2006 ; V icente 2007 ; H arizano v 2019 ). ( 225 c ) is achiev ed via r epr ojection of heads (F anselo w 2003 ; Donati 2006 ; S urán yi 2005 , 2008 ; Georgi and M üller 2010 ). Lastl y , ( 225 d) r epr esents a less con v entional but logicall y possible option, w her e the head is adjoined to a p hrase ( cf. I nternal P air - M erge, M. Richar ds 2009 ). 172 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT (225) a. A head-adjoined position YP Y A sp./M od. Y XP ... < A sp./M od. > ... b. A specifier position XP A sp./M od. X’ ... < A sp./M od. > ... c. A deriv ed head position Asp P /M odP A sp./M od. XP ... < A sp./M od. > ... d. A p hrase-adjoined position XP A sp./M od. XP ... < A sp./M od. > ... One crucial criteria of the landing site is that it must allo w the mo ving head to tak e scope o v er the structurall y lo w er quantificational elements or f ocused elements. This criterion raises a concern to the head-to- head adjunction appr oach in ( 225 a ). If the mo ving head is adjoined to another head, then it cannot c -command the quantificational element in the XP . I t must be assumed that, f or examp le, the scope pr operty of the mo ving head ma y “per colate” to the higher segment of Y such that it can tak e scope ther e. The other thr ee options do not diff er in structural terms and all of them allo w the mo ving head to c -command the comp lement X’/XP . Distinguishing betw een them ma y be a theory - internal issue. I briefl y discuss some potential concerns or r equir ed assumptions in these options. One concern f or the specifier appr oach in ( 225 b ) is that, since the mo ving heads can land abo v e quantificational or f ocused elements of diff er ent types, it does not seem to target a specifier position of a particular (functional) pr ojection ( e.g. F ocus P or T opic P ). M or e importantl y , the examp le in ( 190 ), r epeated belo w in ( 226 ), sho w s that the mo ving heads can target multip le positions, if ther e is mor e than one quantificational/ f ocused element in the same sentence. 173 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT (226) Quantificational adv erbs and subjects and multip le high positions a. f ad v H ai at mui -gaan ev ery -cl hokhaau school g hoi ci begin f sub j daaiboufan most jan person g dou all haau -dou get -able hou good singzik. r esult ‘ A t e v ery school, it begins to be the case that most peop le ar e getting good r esults. ’ ‘ ev ery’ > ‘begin ’ > ‘most’ b. H oi ci begin f ad v hai at mui -gaan ev ery -cl hokhaau school g f sub j daaiboufan most jan person g dou all haau -dou get -able hou good singzik. r esult ‘I t begins to be the case that, at ev ery school, most peop le ar e getting good r esults. ’ ‘begin ’ > ‘ ev ery’ > ‘most’ W hile it is possible that ther e ma y be multip le r elev ant pr ojections in the CP domain, one ma y actuall y need to posit one such pr ojection abo v e each quantificational/ f ocused element that is pr esent in the sentence. In other w or ds, ther e is a corr elation betw een the number of functional pr ojections ( that host the mo ving heads ) in the CP and the number of quantificational/ f ocused elements. Also, mo v e- ment to a specifier position into the CP domain typicall y in v ol v es discourse eff ects, but it is unclear w hat these eff ects w ould be w hen the aspectual v erbs and modal v erbs occup y the high position. 43 As f or the r epr ojection appr oach in ( 225 c ), it is common to assume that a category f eatur e is the trigger of the r epr ojecting mo v ement (F anselo w 2003 ; S urán yi 2005 , 2008 )). S imilar to the head-to- specifier mo v ement appr oach, one w ould be f or ced to posit diff er ent categorial pr ojections in the CP , e.g., diff er ent AspectP s or M odalP s. 43. Indeed, Y .-y . H su ( 2016 , 2019 ) argues that modals mo v e into F ocus P , but she also agr ees that it is not the case that the mo ving head r eceiv es a f ocus interpr etation. This diff ers fr om pr oposals that adopt a head-to-specifier anal y sis, w her e the head in the specifier position r eceiv es a f ocus/ topic interpr etation. See § 4.4.3.2 f or discussions and further arguments against a discourse- f eatur e-driv en appr oach. 174 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT The last option, i.e., the head-to-p hrase adjunction appr oach in ( 225 d), seems to be less pr oblem- atic. I t tak es adv antage of the flexible natur e of adjunction and allo w s diff er ent landing sites f or the pr oposed head mo v ement. As long as the output of the deriv ation is interpr etable, aspectual v erbs and modal v erbs can adjoin fr eel y adjoin to an y syntactic category ( cf. H eim and Kratzer 1998 ). I ackno w ledge that the notion of adjunction as a syntactic operation is still an unsettled issue in the literatur e, but such an appr oach is analogous to other scope-shifting operation such as English Quan- tifier Raising (M a y 1977 , 1985 ; F o x 2000 ), both of w hich ar e able to shift scope and ar e subject to Scope Econom y . 4.4.3.2 The trigger As f or the trigger of the pr oposed head mo v ement, it is not immediatel y clear w hat the trigger of the mo v ement is. I first discuss w hat the trigger is unlik el y to be, and then r eturn to other possible options. W hi le discourse f eatur es or A ’ - f eatur e ar e often held r esponsible f or pr edicate fr onting (V icente 2007 ; Cheng and V icente 2013 ; H ein 2018 ; H arizano v 2019 , among man y others ), it is unlik el y that this is the case f or the pr oposed head mo v ement. F irst of all, the aspectual v erbs and the modal v erbs in the high position do not seem to r eceiv e an y f ocus or topic interpr etation. This is in sharp con- trast with the v erb doubling constructions discussed in Chapter 3. 44 A dditionall y , sentences with an aspectual v erb or a modal v erb in the high position ar e f elicitous in out -of -the-blue contexts or the ‘w hat happened?’ contexts, as illustrated in ( 227 ). (227) An out -of -the-blue/ A ‘w hat happened’ context a. F aatsaang happen me w hat si ev ent aa3? sfp ‘W hat happened?’ 44. R ecall also that modal v erbs r esist a f ocus or topic r eading, as they cannot be targeted in v erb doubling constructions. 175 4.4. P R O POSAL: SCO P E-SHIFTIN G HEAD M O VEMENT b. =( 206b ) H oi ci begin dak onl y Aaming Aaming haau -dou get -able hou good singzik. r esult ‘I t begins to be the case that onl y Aaming is getting go od r esults. ’ I t should be noted that Y .-y . H su ( 2016 , 2019 ) argues that modal mo v ement is an instance of f ocus mo v ement, w her e the modal serv es as a pr opositional f ocus operator and marks the w hole pr oposi - tion/TP as f ocus. She argues f or this claim based on the unacceptability of ( 228 ) ( w hich is the Can- tonese counterpart of the original M andarin examp le ). She suggests that the sentence is ruled out be- cause hoji , as an f ocus operator , interv enes betw een the ( null) question operator (higher in the clause ) and the wh -expr essions, as an instance of ( semantic ) interv ention eff ects Beck ( 1996 , 2006 ). (228) W h -objects and the high position * H oji ma y Aaming Aaming zungji lik e bin-go ? w ho Lit. ‘I t is allo w ed that Aaming lik es w ho?’ H o w ev er , such an anal y sis incorr ectl y pr edicts the unacceptability of ( 198 ), r epeated belo w in ( 229 ), w her e the wh -expr ession in the su bject position. (229) W h -subjects and the high position = ( 198 ) H oji ma y bin-go w ho haa- nin next - y ear heoi go H oenggong? H ong.K ong Lit.: ‘I t is allo w ed that w ho goes to H ong K ong next y ear?’ The subject -object asymmetry f ollo w s fr om the curr ent pr oposal that the high position is onl y licensed if the immediatel y f ollo wing constituent r eceiv es a f ocus interpr etation ( or is a quantifica- tional element ). Thus, ( 228 ) is unacceptable not because hoji causes interv ention eff ects, but because Aaming does not r eceiv e a f ocus interpr etation. 45 45. The intuition that the sentences with a high modal v erb come with pr opositional f ocus or TP f ocus is not imp lau - sible. H o w ev er , instead of suggesting that such an interpr etation is dir ectl y contributed b y the modal mo v ement per se , I 176 4.5. FUR THER E VID EN CE FO R THE P R O POSAL If discourse/ A ’ - f eatur es ar e not r esponsible f or the pr oposed head mo v ement, ther e ar e still tw o possibilities. The first one is to suggest that the mo v ement is trigger ed b y an EP P f eatur e on a null head, without a prior Pr obe-Goal Agr ee r elation (Collins 1997 ; Miy aga w a 2001 ; N evins and Anand 2003 ). An adv antage of this appr oach is that it allo w s us to captur e the lack of scopal ambiguity of sentences lik e ( 227b ). N evins and Anand ( 2003 ) suggest that A - mo v ement onl y f or EP P f eatur e does not r econstruct. If this r easoning can be extended to head mo v ement onl y f or EP P f eatur e, then the lack of scopal ambiguity f ollo w s. Another possibility is to suggest that the pr oposed head mo v ement is not f eatur e-driv en, b ut is trigger ed b y interpr etation or scope considerations, or is “fr ee, ” in a sense similar to English Quantifier Raising and J apanese/ German A -scrambling. (M a y 1977 ; 1985 ; F o x 2000 ; Bobaljik and W urmbrand 2012 , i.a.). U nder the model pr oposed b y Bobaljik and W urmbrand ( 2012 ), the lack of r econstruction of this mo v ement is due to the “¾ S ignatur e Eff ects” , in a w a y similar to A -scrambling in J apanese and German, w hich cannot be r econstructed f or scope interpr etation. The pr oposed head mo v ement is compatible with these possibilities. 46 4.5 F urther e vi dence f or the pr oposal In this section, I discuss further evidence f or the pr oposed mo v ement. In § 4.5.1 , I examine cases w her e multip le quantificational heads ar e stack ed in the high position. In § 4.5.2 , I sho w that it is constrained b y a locality constraint on mo v ement, Shortest M o v e. In 4.5.3 , I r eport cases w her e the quantifica- tional heads can undergo A cr oss-the- Boar d (A TB ) mo v ement. All these cases fall out naturall y fr om the pr oposed mo v ement anal y ses. Lastl y , in § 4.5.4 , I examine w hether the pr oposed mo v ement ex - hibits island eff ects, w hich ar e typical pr operties of syntactic mo v ement. I suggest that ther e is no conclusiv e evidence f or the pr esence or absence of island eff ects. suggest that it is a r esult of the change of the calculation of the f ocus sets/ f ocus alternativ es, as suggested in § 4.2.3 . M odal mo v ement allo w s the modal v erbs to escape fr om the f ocus scope of the f ocus operator , lea ving the r est of the sentence, usuall y a TP , to be within the f ocus set. This ma y giv e rise to the intuition that the TP as a w hole is f ocused. 46. The discussion in the subsection has gr eatl y benefited fr om discussions with S tefan K eine and R oum y ana P anchev a. 177 4.5. FUR THER E VID EN CE FO R THE P R O POSAL 4.5.1 S tacking of quantifi cati onal heads in the high positi on I t is suggested that a quantificational element can license the pr oposed head mo v ement, as mo ving acr oss it potentiall y leads to scope eff ects (hence obeying Scope Econom y ). I t is also suggested that the mo ving head is quantificational b y natur e (i.e., a generalized quantifier ). These combine to pr edict that the mo v ement of a quantificational head to the high position w ould license the mo v ement of another quantificational head to the high position. In other w or ds, it is pr edicted that quantificational heads can be “ stack ed” in the high position, if the pr oposed mo v ement app lies mor e than once. This pr ediction is borne out. Observ e that hoici ‘begin ’ and hoji ‘ma y’ can co-occur in the (post - subject ) lo w positions in ( 230a ). S ince the subject is quantificational, the mo v ement of hoici ‘begin ’ to a higher position is possible, as it obey s Scope Econom y ( as in ( 230b ). F urthermor e, the mo v ement of hoji ‘ma y’ to a position higher than hoici ‘begin ’ is also allo w ed, as illustrated in ( 230c ). H oji ‘ma y’ tak es scope o v er both hoici ‘begin ’ and dak ‘ onl y’ . (230 ) S tacking of aspectual v erbs and deontic modals a. ‘ onl y’ > ‘begin ’ > ‘ma y’ Dak onl y Aaming Aaming hoi ci begin hoji ma y zou earl y fan. sleep ‘ Aaming begins to be allo w ed to sleep earl y . ’ b. ‘begin ’ > ‘ onl y’ > ‘ma y’ H oi ci begin dak onl y Aaming Aaming hoji ma y zou earl y fan. sleep ‘I t begins to be the case that it is allo w ed that onl y Aaming sl eeps earl y . ’ c. ‘ma y’ > ‘be gin ’ > ‘ onl y’ H oji ma y hoi ci begin dak onl y Aaming Aaming zou earl y fan. sleep ‘I t is allo w ed that it begins to be the case that onl y Aaming sl eeps earl y . ’ T o see an additional examp les with diff er ent quantificational heads, consider the sentences in ( 231 ). The modal v erb wui ‘will’ and the aspectual v erb gaizuk ‘ continue’ can be stack ed in the high 178 4.5. FUR THER E VID EN CE FO R THE P R O POSAL position (in the embedded clause ), as sho wn in ( 231c ). (231) S tacking of futur e modals and aspectual v erbs a. ‘ onl y’ > ‘will’ > ‘ continue’ N go I gok dak think [dak onl y gupiu stock wui will gaizuk continue sing]. rise ‘I think that onl y stock (prices ) will continue to rise. ’ b. ‘ continue’ > ‘ onl y’ > ‘will’ N go I gok dak think [ gaizuk continue dak onl y gupiu stock wui will sing]. rise ‘I t hink that it will continue to be the case that onl y stock (prices ) rise. ’ c. ‘will’ > ‘ continue’ > ‘ onl y’ N go I gok dak think [ wui will gaizuk continue dak onl y gupiu stock sing]. rise ‘I t hink that it will continue to be the case that onl y stock (prices ) rise. ’ U nder the curr ent pr oposal, the deriv ation of ( 230c ), f or instance, in v ol v es tw o steps, namel y , the mo v ement of hoji ‘ma y’ , f ollo w ed b y the mo v ement of hoici ‘begin ’ . N ote that the tw o mo v ement steps ar e independent of each other (i.e., they need not occur together ), and both observ e Scope Econom y . (232) A tw o-step deriv ation of ( 230c ) ma y begin [ TP onl y Aaming i [ v P t i sleep-earl y ]] 4.5.2 Shortest M o v e In addition to Scope Econom y , the pr oposed head mo v ement is also subject to a locality constraint on mo v ement, Shortest M ove . Consider the configuration in ( 233 ) w her e ther e ar e a quantificational subject and a non-quantificational adv erbial in a sentence. (233) A configuration sho wing possible landing sites under Scope Econom y · [ TP A dv erbial [-Q ] ¶ [ TP S ubject [+Q ] A sp./M od. [ v P ... ] ] ] 179 4.5. FUR THER E VID EN CE FO R THE P R O POSAL I t has been sho wn that the position right abo v e the quantificational subject (= ¶ ) is a possible landing site in adher ence to Scope Econom y . The position right abo v e the non-quantificational adv erbial (=· ), at least in princip le, should be another possible landing site, since the mo ving heads can tak e wide scope fr om ther e o v er the quantificational subject. This mo v ement step does not violate Scope Econom y and should be allo w ed. This, ho w ev er , is not the case, as illustrated in ( 234 ). 47 (234) A non-quantificational adv erbial and a quantificational subject · ?? H oi ci begin [ ad v hai at ngodei our hokhaau school ] ¶ hoi ci begin [ sub j daaiboufan most jan person ] dou all haau -dou get -able hou good singzik. r esult ‘ A t our school, it begins to be the case that,most peop le ar e getting good r esults. ’ The una v ailability of the high position in · suggests that the mo v ement of hoici cannot be longer /higher than it needs to shift scope. I t must land right above the quantificational subject. S uch a r estriction does n ot fall out fr om Scope Econom y . I su ggest that this f ollo w s fr om a v ersion of Shortest M o v e, giv en in ( 235 ). R ecall that I ha v e suggested that Scope Econom y is a syntactic, deriv ational constraint on mo v ement, and that certain f ormal logical pr operties ar e accessible in syntax. I suggest that the pr oposed Shortest M o v e in ( 235 ) bears a similar character in the sense that it app lies right after the ev aluation of Scope Econom y , and dictates that the time w hen a scope-shifting operation is sanctioned b y Scope Econom y is the onl y possible time that the scope-shifting operation can app l y . 48 47. The high position in · is a v ailable if the adv erbial is quantificational, as sho wn in ( 190 ). 48. This is r eminiscent of the “ A djoin-as-soon-as-possible” condition pr oposed in Erlewine ( 2015 , 2017 ), w hich is orig - inall y pr oposed to r egulate adjunction of Externall y - M erged elements. If this can be extended to Internall y - M erged ele- ments, and if the the pr oposed mo v ement is achiev ed via adjunction ( as one of the f our possibilities discussed in § 4.4.3.2 ), then the pr oposed Shortest M o v e ma y be subsumed under the “ A djoin-as-soon-as-possible” condition. 180 4.5. FUR THER E VID EN CE FO R THE P R O POSAL (235) Shortest ( scope-shifting) M o v e A scope-shifting operation must mo v e a quantifier to the closest position in w hich it shifts scope . As such, ( 235 ) dictates that the position ¶ in ( 234 ) is the onl y possible landing site that deliv ers the wide ‘begin ’ -scope, as it is the closest position. This exp lains w h y the landing site of the mo ving quan- tificational heads must be immediately f ollo w ed b y a quantificational/ f ocused element ( as stated in ( 202 )), as it is the closest position. I tak e this as further evidence f or a mo v ement appr oach to the high position of quantificational heads. Shortest M o v e as f ormulated in ( 235 ) pr o vides an exp lanation of the unacceptable cases discussed at the end of § 4.4.2.2 , r epeated belo w as ( 236 ). (236) A puzzle on the unacceptability of ( 195 b ) a. =( 195 b ) * H oji ma y Aaming Aaming hai co p heoi go H oenggong. H ong.K ong Int.: ‘I t ma y be the case that Aaming goes to H ong K ong. ’ b. baseline of ( 195 b ) Aaming Aaming hai co p hoji ma y heoi go H oenggong. H ong.K ong ‘ Aaming ma y go to H ong K ong. ’ The sentence in ( 236 a ) appears to be pr oblematic: if it has the base structur e in ( 236 b ), then the mo v ement of hoji ‘ma y’ to the high position should not violate Scope Econom y , since it cr osses the f ocus operator ( and aff ects the calculation of the f ocus sets ). This suggests that ( 236 a ) is disallo w ed f or r easons other than Scope Econom y . W ith ( 235 ), the unacceptability of ( 236 a ) can be attributed to r easons similar to w h y ( 234 ) is ruled out: it does not land right abo v e the f ocus operator , i.e., its mo v ement is too “long” . If the mo v ement does not cr oss the subject, it is allo w ed, as sho wn in ( 237 ). 181 4.5. FUR THER E VID EN CE FO R THE P R O POSAL (237) M o v ement of hoji ‘modal’ in comp liance to Shortest M o v e and Scope E conom y Aaming Aaming hoji ma y hai co p heoi go H oenggong. H ong.K ong ‘ Aaming ma y go to H ong K ong. ’ The spirit behind ( 235 ) is similar but not identical to F o x’ s v ersion suggested in F o x ( 2000 , p.23). (238) Shortest (interpr etable ) M o v e (F o x’ s v ersion ) [ Quantifier Raising] must mo v e a [quantifier p hrase ] to the closest position in w hich it is inter - pr etable . In other w or ds, a [quantifier p hrase ] must al w a y s mo v e to the closest clause-denoting element that dominates it. The crucial diff er ence w ith ( 235 ) lies in w hat is r elev ant in the calculation of “ closest position ” . F o x’ s v ersion suggests that it is the clause-denoting elements (i.e., the nodes that ar e of pr oposition type t ). Importantl y , mo v ement to all these positions is constrained b y Scope Econom y , i.e., mo ving to each of these positions must shift scope r elations. On the other hand, the v ersion pr oposed her e is less r estricted in that it suggests that onl y scope-taking elements count w hen calculating “the closest position ” . Eff ectiv el y , the tw o v ersions mak e diff er ent pr edictions on w hether the mo ving quantifier element can skip a clause-denoting non-quantificational node. (239) Diff er ent pr edictions on the mo v ement of ( w her e XP /YP /ZP ar e all clause-denoting) a. F o x’ s Shortest M o v e: * [ XP [+Q ] ... [ YP [-Q ] ... [ ZP < >... ] ] ] YP = non-skippable b. The pr oposed Shortest M o v e: O K [ XP [+Q ] ... [ YP [-Q ] ... [ ZP < >... ] ] ] YP = skippable The pr ediction made b y the pr oposed v ersion is borne out, at least in Cantonese, since the mo v e- ment of ‘begin ’ can skip the TP . 49 49. The motiv ation f or F o x’ s v ersion comes fr om the contrast in (i), discussed in F o x ( 2000 ), p.64, fn.52. F o x’ s v ersion deriv es the contrast without r el ying on the assumption that Q R is clause-bounded ( contr a M oltmann and S zabolcsi 1994 ). This is because in (ia ) the CP is the closest clause-denoting node f or the univ ersal quantifier , but it is non-quantificational, and the position is thus ruled out b y Scope Econom y . In contrast, in (ib ) the CP is quantificational ( due to the wh - expr ession ) and thus the position is sanctioned. T ogether with the assumption that know is quantificational, the univ ersal quantifier mo v es further to the matrix v /VP and then to TP o v er the existential subject, deliv ering the wide scope r eading. 182 4.5. FUR THER E VID EN CE FO R THE P R O POSAL (240 ) The pr oposed mo v ement ma y skip interv ening nodes that denote clauses = ( 188 a ) H oi ci begin [ T o p P [ cyunbou ev ery jan ] person [ TP Aaming Aaming dou all hou v ery jansoeng] praise ]. ‘I t begins to be the case that Aaming praises ev ery one. ’ Concerning the status of Shortest M o v e in ( 235 ), it might be subsumed under Scope Econom y since both mak e r ef er ence specificall y to scope and ar e ev aluated locall y . E vidence f or this claim w ould r el y on o v ert scope-shifting operations that ar e subject to Scope Econom y . As far as I am a w ar e, ther e ar e tw o such cases: det erminer raising in J apanese (T akahashi 2002 ) and negativ e auxil - iary in v ersion in some v arieties of English (M atyiku 2017 ). Shortest M o v e in ( 235 ) is compatible with their data, w her e the mo ving heads land right abo v e the element that they scope o v er . Alternativ el y , Shortest M o v e in ( 235 ) might be subsumed under Minimal Link Condition (Chomsky 1995b ), Short - est A ttract/M o v e (N . Richar ds 2001 ), or the f eatural R elativized Minimality (Rizzi 2001 , 2004 ), if the r elev ant f ormal logical pr operties ar e r ealized as syntactic f eatur es ( as discussed in § 4.4.1.2 ), such that the notion of “ shortest” is calculated based on the distance betw een the mo v ement trigger and the corr esponding syntactic f eatur es. H o w ev er , additional assumptions is needed to constrain the land- ing site of the mo v ement, rather than the launching site (i.e., the base position of t he mo ving element ). I lea v e the pr ecise f ormulation open. 4.5.3 M o v ement out of coor dinate structur es One further piece of evidence f or the pr oposed mo v ement comes fr om A cr oss-the- Boar d (A TB ) mo v e- ment out of coor dinate structur es. In Cantonese, disjunctiv e scope can be mark ed b y m-hai ... zauhai (i) a. *W ide univ ersal scope: One girl kno w s [ CP that [-Q ] ev ery bo y bought a pr esent f or M ary ]. b. O K W ide univ ersal scope: One girl kno w s [ CP w hat [+Q ] ev ery bo y bought f or M ary ]. H o w ev er , as F o x admits, the una v ailability of wide univ ersal scope in (ii) ( discussed in M oltmann and S zabolcsi 1994 ) is unexp lained. I lea v e open the pr ecise f ormulation of the locality constraint on English Q R. (ii) *W ide univ ersal scope: One girl w onder [ CP w hat [+Q ] ev ery bo y bought f or M ary ]. 183 4.5. FUR THER E VID EN CE FO R THE P R O POSAL ... ‘ either ... or ... ’ . In ( 241 ), the tw o disjuncts ar e sentential and both contain the aspectual v erb hoici ‘begin ’ . (241) A coor dinate structur e containing hoici ‘begin ’ in each clause Context: I t is hear d that either Aaming or Aaf an has studied very har d r ecently. The speaker guesses: M - hai not -co p Aaming Aaming hoi ci begin haau -dak get -resul t hou, good, zau - hai then-co p Aafan Aafan hoi ci begin haau -dak get -resul t hou. good ‘Either Aaming begins to perf orm w ell, or Aafan begins to perf orm w ell. ’ Importantl y , it is possible f or hoici ‘begin ’ to occup y the high position abo v e the disjunctiv e scope, as in ( 242 ). The scope of hoici ‘begin ’ is extended and no w it scopes o v er the disjunctiv e scope. The sentence is true under a diff er ent scenario. N ote that it is not true under the scenario in ( 241 ). (242) A TB mo v ement out of coor dinate structur es Context: I n past ex ams, either Aaming or Aaf an or Chris got good r esults. R ecently, Chris is dis - tr acted fr om study, and now either Aaming is getting good r esults, or Aaf an is getting good r esults. H oi ci begin m- hai not -co p Aaming Aaming haau -dak get -res hou, good, zauhai then-co p Aafan Aafan haau -dak get -res hou. good ‘I t begins to be the case that either Aaming perf orms w ell, or Aafan perf orms w ell. ’ The high position can be deriv ed b y app l ying the pr oposed head mo v ement in an acr oss-the-boar d (A TB ) fashion (R oss 1967 ; W illiams 1978 ). Specificall y , the aspectual v erb hoici ‘begin ’ in each disjunct undergoes head mo v ement to a clause-external, high position, in a w a y depicted in ( 243 ). 50 (243) A simp lified r epr esentation of the deriv ation of ( 242 ), in English gloss begin [ [ CP1 either Aaming ... ] [ CP2 or Aafan ... ] ] A TB mo v ement 50. A TB mo v ement in attested elsew her e in Chinese, such as in gapping - lik e constructions (T ang 2001 ) and wh - mo v ement (P an 2011 ). 184 4.5. FUR THER E VID EN CE FO R THE P R O POSAL N ote that such mo v ement is impossible if onl y one instance of hoici ‘begin ’ mo v es out. This is be- cause the mo v ement w ould violate the Coor dinate S tructur e Constraint, w hich pr ecludes mo v ement out of coor dinate structur e (R oss 1967 ). 51 (244) Illicit non- A TB mo v ement out of coor dinate structur es * H oi ci begin m- hai not -co p Aaming Aaming hoici begin haau -dak get -res hou, good zauhai then-co p Aafan Aafan haau -dak get -res hou. good Int.: ‘I t begins to be the case that either Aaming begins to perf orm w ell, or Aafan perf orms w ell. ’ 4.5.4 A r emar k on the indeterminacy of island sensiti vity In this last subsection, I examine w hether the pr oposed head mo v ement exhibits island eff ects, since island sensitivity is commonl y tak en to support syntactic mo v ement. 52 H o w ev er , it is sho wn, f or independent r easons, that ther e is no clear evidence fr om syntactic islands f or or against a mo v ement appr oach. This is because it is difficult, if not impossible, to fix the base position of the aspectual v erbs and modal v erbs in bi -clausal structur es. In or der to determine w hether the pr oposed head mo v ement is sensitiv e to syntactic islands, the configuration containing a comp lex NP island in ( 245b ) is of inter e st. (245) Comp lex NP I sland interv ention a. Baseline: S V [ Comp lex NP [ S A sp./M od. VP ] NP ] b. T arget: A sp./M od. S V [ Comp lex NP [ S VP ] NP ] The f ollo wing sentences illustrate these configurations with hoici ‘begin ’ . N ote that the subjects in both 51. The sentence in ( 244 ) without the initial hoici ‘begin ’ is acceptable under the f ollo wing context: if Aaming is a y ounger br other and Aafan is his sister , and Dad bought just one pr esent, sa ying, “if Aaming is doing better than last time, then he gets the pr esent. If Aaming didn ’t and Aafan is doing as good as last time, then Aafan gets the pr esent. ” One da y , the pr esent w as gone, and the speak er inf err ed fr om this b y sa ying ( 244 ) without the initial hoici . 52. I thank Colin Da vis f or critical and constructiv e comments on this issue. 185 4.5. FUR THER E VID EN CE FO R THE P R O POSAL matrix and embedded must be quantificational in or der to license the pr oposed head mo v ement. (246) T w o positions of hoici ‘begin ’ in sentences w ith a comp lex NP island a. Dak onl y Aaming Aaming zung still soengseon believ e [ NP [go-go cl-cl dou all hoi ci begin haau -dak get -able hou ] good ge ge siusik]. rumor ‘Onl y Aaming still believ es the rumor that ev ery one begins to perf orm w ell. ’ b. H oi ci begin dak onl y Aaming Aaming zung still soengseon believ e [ NP [go-go cl-cl dou all haau -dak get -able hou ] good ge ge siusik]. rumor ‘I t begins that onl y Aaming still believ es the rumor that ev ery one perf orms w ell. ’ The high position of hoici ‘begin ’ in ( 246 b ) suggests that mo v ement occurs and its acceptability indicates that ther e is no island violation. One ma y then suggest that the pr oposed head mo v ement is insensitive to syntactic islands, but this is based on the assumption that ( 246 b ) is deriv ed fr om ( 246 a ), w her e hoici originates fr om the embedded clause. (247) A possible deriv ation f or ( 246 b ), mo v ement acr oss a comp lex NP island begin S V [ Comp lex NP [ S VP ] NP ] H o w ev er , ther e could be another possible deriv ation f or ( 246 b ), w her e hoici is base generated in the matrix clause. (248) An alternativ e deriv ation f or ( 246 b, no mo v ement acr oss a comp lex NP island) begin S V [ Comp lex NP [ S VP ] NP ] S ince the target configuration ma y ha v e either one of the possible base structur es, it is crucial to de- termine the base position of hoici ‘begin ’ . H o w ev er , ther e ar e at least tw o r easons that the base posi - tions cannot be easil y determined. On one hand, since aspectual v erbs and modal v erbs ar e raising pr edicates as suggested (Y .-H. A. Li 1990 ; Lin and T ang 1995 , i.a.), they do not impose selectional r e- 186 4.6. AL TERN A TIVE AN AL Y SES T O A HEAD M O VEMENT AP P R O A CH quir ement on the subject (i.e., they onl y select a clausal comp lement ). Ther ef or e, both the matrix and the embedded positions ar e possible base position of these v erbs. On the other hand, r econstruction is una v ailable f or independent r easons (i.e., arguabl y due to Scope Econom y; see § 4.4.3.2 ). I t is thus not possible to detect the base positions of these v erbs based on interpr etations either . S imilar is- sues arise in other island configurations. As such, I conclude that standar d island diagnostics do not pr o vide evidence f or or against a mo v ement appr oach to the high position. 53 I t should be ackno w ledged that the pr oposed head mo v ement does not exhibit typical mo v ement pr operties such as r econstruction, scope ambiguity and island eff ects. The evidence f or mo v ement must r el y on other less standar d (but not necessaril y less con vincing) evidence discussed in pr evious subsections. 4.6 Alternati v e anal y ses to a head mo v ement appr oach In this section, I argue against f our alternativ e anal y ses to the pr oposed head mo v ement account. I first discuss tw o non- mo v ement appr oaches to the positional alternation of aspectual v erbs and modal v erbs in § 4.6.1 and § 4.6.2 . I then discuss tw o v ariants of a p hrasal mo v ement appr oach in § 4.6.3 and in § 4.6.4 , , w her e mo v ement is in v ol v ed, but it is p hrasal mo v ement instead of head mo v ement. 53. S imilar issues arise w hen determining w hether the pr oposed head mo v ement is clause-bounded or not, i.e., w hether a head can exit an embedded clause ( and r esult in a long -distance dependency ). The r elev ant configuration is schematicall y illustrated in (i) belo w . Ther ef or e, ther e is no clear evidence f or or against the clause boundedness of the pr oposed head mo v ement either . (i) CP interv ention a. T arget: A sp./M od. S V [ CP S VP ] b. P ossible base structur e 1 : S V [ CP S A sp./M od. VP ] c. P ossible base structur e 2: S A sp./M od. V [ CP S VP ] 187 4.6. AL TERN A TIVE AN AL Y SES T O A HEAD M O VEMENT AP P R O A CH 4.6.1 M ultip l e base positi ons of aspectual v erbs and modal v erbs U nder a non- mo v ement appr oach, it might be suggested that the high and lo w positions of aspectual v erbs and modal v erbs ar e not deriv ationall y r elated b y mo v ement. Instead, they w ould be merged dir ectl y in the high or lo w positions (i.e., they ar e adjunct - lik e elements ). R ecall the paradigm discussed in § 4.3.1 , simp lified in ( 249 ). (249) A simp lified r epr esentation of the sentences in ( 175 ) and ( 176 ) a. Aaming > begin > VP cf. ( 175a ) b. * Begin > Aaming > VP cf. ( 175b ) c. Onl y Aaming > begin > VP cf. ( 176a ) d. Begin > Onl y Aaming > VP cf. ( 176b ) S uch a multip le base position appr oach could captur e the admissible cases in ( 249 a, c, d). In or der to rule out cases lik e ( 249 b ), this appr oach w ould ha v e to emp lo y an additional constraint w hich dic - tates that the high base position is onl y a v ailable w hen the f ollo wing constituent is quantificational or r eceiv es a f ocus interpr etation. A t least tw o issues arise with such an appr oach. F irst, this appr oach must look ahead to determine the timing of adjunction of the r elev ant heads. This is because if ther e is a quantificational element that is going to be merged, these heads must “w ait” until then. In other w or ds, the timing of adjunction of these heads must be based on the the r esult of this look -ahead operation. In contrast to this, a head mo v ement account coup led with Scope Econom y does not giv e rise to similar look -ahead issues. Second, the constraint on the high position appears to be a r e-statement of the generalization in ( 202 ), or w e must stipulate a base-position v ersion of Scope Econom y , w hich dictates “ do not merge high unless y ou tak e wider scope ther e” . S uch a constraint on base position is, as far as I kno w , rar el y hear d of. I t is unclear w h y merging at a lo w position w ould be mor e pr ef err ed to merging at a high one. 188 4.6. AL TERN A TIVE AN AL Y SES T O A HEAD M O VEMENT AP P R O A CH One possibility is to suggest that the high position of aspectual v erbs and modal v erbs ma y impose a selectional r equir ement on the adjacent XP . F or examp le, the unacceptability of ( 249 b ) f ollo w s if w e assume that hoici ( semanticall y ) selects a quantificational element and thus a pr oper name in the subject position fails this r equir ement. S uch a r equir ement, at least superficiall y , bears similarity to English almost , illustrated belo w . (250 ) The selectional r equir ement of almost in English a. J ohn/T en peop le almost died. cf. ( 249 a,c ) b. * [ Almost J ohn ] died. cf. ( 249 b ) c. O K [ Almost ten peop le ] died. cf. ( 249 d) H o w ev er , ther e ar e tw o issues concerning this suggestion. F irst, if it is assumed that hoici ‘begin ’ imposes a selectional r equir ement on the adjacent element, it must be also assumed that the comp le- ment clause selected b y hoici ‘begin ’ ( such as the VP in ( 307 a, c )) is also quantificational, but it is unclear w h y this is so. Second, a selectiona l r equir ement pr esupposes constituency betw een the selecting element and the selected element. The sentence in ( 251 ) sho w s that the univ ersal quantifiers do not f orm a con- stituent with hoici ‘begin ’ . N ote that the co v erb tung ‘with ’ tak es nominal structur es as its comp lement. (251) H oici ‘begin ’ does not f orm a constituent with the nominal quantifiers * N go I tung with hoi ci begin [ mui -go ev ery -cl jan ] person dou all gong - zo talk ed je. thing Int.: ‘I t begins to be the case that I ha v e talk ed to ev ery one. ’ N otabl y , this is in contrast to caa-m-do ‘ almost’ in Cantonese w hich sho w s similar distributional pr op- erties to English almost . I t f orms a constituent with the adjacent nominal. 189 4.6. AL TERN A TIVE AN AL Y SES T O A HEAD M O VEMENT AP P R O A CH (252) N go I tung with caa- m -do almost [ mui -go ev ery -cl jan ] person dou all gong - zo talk ed je. thing ‘I ha v e talk ed to almost ev ery one. ’ I ther ef or e suggest that the una v ailability of the high position of aspectual v erbs and modal v erbs cannot be attributed to selectional r equir ements. I conclude further that the base position appr oach does not adequatel y captur e the paradigm in ( 307 ). 4.6.2 An in -situ appr oach to aspectual v erbs and modal v erbs Another v ariant of a non- mo v ement appr oach is to suggest that the quantificational heads under dis- cussion ar e indeed in-situ, and the w or d or der alternation is due to the optional mo v ement of XP s surr ounding them ( e.g., subjects, topics, adv erbials, etc.), as illustrated in ( 253 ). A conceiv able possi - bility of this XP mo v ement in Cantonese might be topic mo v ement, w hich can target diff er ent types of constituents. U nder this appr oach, no head mo v ement is needed. (253) A h ypothetical appr oach that emp lo y s optional XP mo v ements (XP ) [ AspectP /M odalP A sp./M od. [ (XP ) ... ] ] Optional XP mo v ement Ther e ar e, ho w ev er , tw o r easons not to adopt such an anal y sis. F irst, r ecall the paradigm in ( 307 ), r epeated in ( 254 ). W hile this appr oach w ould be able to captur e the w or d or der alternation in ( 254 c -d), it wr ongl y pr edicts that ( 254 b ) to be acceptable. (254) A simp lified r epr esentation of the sentences in ( 175 ) and ( 176 ) = ( 307 ) a. Aaming > begin > VP cf. ( 175a ) b. * Begin > Aaming > VP cf. ( 175b ) c. Onl y Aaming > begin > VP cf. ( 176a ) 190 4.6. AL TERN A TIVE AN AL Y SES T O A HEAD M O VEMENT AP P R O A CH d. Begin > Onl y Aaming > VP cf. ( 176b ) In other w or ds, this appr oach does not captur e the sensitivity to the quantificational/ f ocus natur e of XP . In or der to rule out examp les lik e ( 254 b ), it must be assumed that non-quantificational ele- ments must mo v e acr oss the aspectual v erbs and modal v erbs ( w hen embedded under them ), w her eas quantificational elements optionall y do so. H o w ev er , this assumption w ould also r equir e all non- quantificational objects ( e.g., hou singzik ‘ good r esult’), to mo v e, contrary to facts. R easoning along this line w ould also need to account f or w h y non-quantificational/ non- f ocused elements ar e special in that they must undergo mo v ement in the r elev ant configurations, or w h y quan- tificational/ f ocused elements ar e ex ceptionall y allo w ed to r esist mo v ement. S uch an account is not impossible to f ormulate, but w ould be less p lausible w hen compar ed to a head mo v ement account, since the latter can r esort to Scope Econom y to captur e the sensitivity to quantificational/ f ocused elements, and Scope Econom y r eceiv es independent support fr om diff er ent p henomena in diff er ent languages (F o x 2000 ; T akahashi 2002 ; M atyiku 2017 , i .a.). 4.6.3 A r emnant mo v ement appr oach Another alternativ e anal y sis to a head mo v ement appr oach is a r emnant mo v ement appr oach. I t shar es the idea that mo v ement of aspectual v erbs and modal v erbs is in v ol v ed, but diff ers fr om the curr ent pr oposal in that the mo ving element is not a head, but a p hrase. S uch an appr oach w ould suggest that w hat surfaces as head mo v ement could indeed be r emnant VP mo v ement in disguise (den Besten and W ebelhuth 1990 ; K oopman and S zabolcsi 2000 ; M ahajan 2003 , among man y others ). Specificall y , it might be suggested that, bef or e VP mo v ement, all other elements ex cept the v erb ar e extracted fr om the VP . As a r esult, the r emnant VP w ould contain just a v erb, and w hen the VP mo v es, it appears that the v erb is mo ving on its o wn, but in fact this is an instance of p hrasal mo v ement. Imp lementing this idea on ( 176b / 206b ), illustrated with English glosses belo w , the v P comp lement of hoici ‘begin ’ is first extracted to a higher position, as indicated in the step in ( 255b ). S ubsequentl y , AspectP containing onl y the aspectual v erb is fr onted, as in ( 255c ). 191 4.6. AL TERN A TIVE AN AL Y SES T O A HEAD M O VEMENT AP P R O A CH (255) Deriving ( 176b fr om ( 176a ) und er a r emnant mo v ement appr oach a. Base structur e: [ TP onl y Aaming [ v P1 begin [ v P2 get -good- r esult ] ] ] b. Fr onting o f the complement o f ‘begin’ , i.e. v P: [ TP onl y Aaming [ v P get -good- r esult ] [ AspectP begin t v P ] ] v P mo v ement c. R emnant phr asal movement o f Aspect P: [ AspectP begin t v P ] [ TP onl y Aaming [ v P get -good- r esult ] t AspectP ] ( r emnant ) AspectP mo v ement Ther e ar e, ho w ev er , tw o issues with this appr oach. F irst, the legitimacy of fr onting v P in ( 255b ) must be stipulated, since this intermediate step b y itself does not f orm an acceptable sentence, as sho wn in ( 256 ). In other w or ds, its app lication is dependent on the subsequent AspectP mo v ement, i.e., v P mo v ement is construction-specific. (256) F r onting of v P abo v e AspectP is disallo w ed * Dak onl y Aaming Aaming [ v P haau -dou get -able hou good singzik] r esult hoi ci . begin ‘Onl y Aaming is such that he begins t o get good r esults. ’ Secondl y , and cruciall y , w hile this appr oach deriv es the desirable w or d or der , it does not deliv er the r elev ant scope facts. A ccor ding to the structur e in ( 255c ), ‘begin ’ is “buried” in the AspectP , and it does not c -command dak ‘ onl y’ . A surface scope r eading w ould r equir e some non-standar d scope- taking mechanism or r edefinition of the notion of c -command. A dditionall y , ev en if hoici ‘begin ’ could tak e scope fr om within the AspectP , the sentence w ould be expected to be ambiguous, since v P fr onting r econstructs ( cf. H uang 1993 ; assuming the same f or AspectP fr onting). H o w ev er , hoici ‘begin ’ unambiguousl y tak es wide scope in the high position. As such, it is unlik el y that p hrasal mo v e- 192 4.6. AL TERN A TIVE AN AL Y SES T O A HEAD M O VEMENT AP P R O A CH ment is at p la y her e; instead, a head mo v ement account straightf orw ar dl y accounts f or both the sur - face w or d or der and the wide scope r eading of ‘begin ’ . 4.6.4 M o v ement of aspectual v erbs and modal v erbs as p hr asal mo v ement Another v ariant of the p hrasal mo v ement appr oach is to suggest that aspectual v erbs and deontic/ futur e modals ar e p hrases, instead of heads (i.e., they ar e not minimal elements ). If so, the pr oposed mo v e- ment should be r egar ded as p hrasal mo v ement, instead of head mo v ement. I t should be noted that the head v s. p hrase distinction is not al w a y s clear in languages lik e Chinese w hich lack v erbal inflection. The p hrase-structural status of an element can onl y be diagnosed indi - r ectl y . H er e, I first off er an argument fr om VP ellipsis f or the head status of deontic/ futur e modals ( a nd I will r eturn to aspectual v erbs belo w ). I t has been observ ed that a VP can be elided i n M andarin w hen it f ollo w s modal auxiliaries (T sai 2015 ; La w and N da yiragije 2017 ), similar to English (Sag 1976 , i.a.). The same app lies to Cantonese, illustrated belo w (∆ marks the elided site ). (257) M odals that license VP ellipsis Aaming Aaming hoji / wui ma y / will [lai ], come Aafan Aafan dou also hoji / wui ma y / will ∆ . ‘ Aaming ma y / will come, and Aafan ma y / will, too. ’ This is in contrast with other p hrasal/ adv erbial modal elements, w hich do not license VP ellipsis. (258) M odals that do not license VP ellipsis ?? Aaming Aaming bitseoi necessaril y [lai ], come Aafan Aafan dou also bitseoi necessaril y ∆ . ‘ Aaming must come, Aafan must, too. ’ The contrast f ollo w s if w e mak e the common assumption that VP ellipsis r equir es head licensing ( or head go v ernment, H uang 1993 ; T sai 2015 ). Cruciall y , the modals that can license VP ellipsis ar e those that can undergo the pr oposed mo v ement, suggesting their head status. 193 4.7. D ISCUSSI O NS AND IMP LI CA TI O NS The situation concerning aspectual v erbs is diff er ent, ho w ev er . F or examp le, hoici ‘begin ’ does not license VP ellipsis. (259) Aspectual v erbs do not license VP ellipsis ?? N i -po this-cl faa flo w er hoi ci begin [ maanmaan slo w l y bin turn hung] r ed laa3, sfp go-po that -cl faa flo w er dou also hoi ci begin ∆ laa3. sfp Int.: ‘The flo w er begins to turn r ed slo w l y . That flo w er begins, too ’ This might be tak en to suggest that aspectual v erbs ar e indeed adv erbs (hence p hrasal elements ). Anticipating the discussions in § 4.7.2 , the pr oposed mo v ement also app lies to p hrases, w hich can be extended to capt ur e distribution of a subset of adv erbs. The p hrasal/ adv erb status of aspectual v erbs w ould not be a concern to the pr oposal, but this w ould mean that the evidence f or head mo v ement with semantic eff ects come ex clusiv el y fr om modal v erbs. I t might also be that the unacceptability of ( 259 ) is due to the fact that VP ellipsis is additionall y constrained b y other princip les, and being a head is onl y a necessary but not sufficient condition in licensing VP ellipsis. In other w or ds, VP ellipsis is a unidir ectional diagnostic: if an element can license VP ellipsis, then it is a head. But it is agnostic on elements that fail to license VP ellipsis. If this is the case, ( 259 ) does not necessaril y speak against the head status of aspectual v erbs. Indeed, the possibility that VP ellipsis r equir es something mor e than head licensing is r ecentl y exp lor ed in Lee and P an ( 2021 ). 4.7 Discussi ons and imp li cati ons 4.7 .1 Semanti c eff ects of head mo v ement The pr oposed mo v ement of aspectual v erbs and ( a subset of ) modal v erbs in Cantonese constitute new evidence f or head mo v ement with scope eff ects, alongside the cases discussed in § 4.2 . H er e it can be noted that the case in Cantonese a v oids certain concerns facing head mo v ement with scope eff ects. 194 4.7. D ISCUSSI O NS AND IMP LI CA TI O NS The first of these concerns generality . Among the f our types of quantificational heads, namel y , de- terminers, negation, modal v erbs, and aspectual v erbs ( as discussed in § 4.2.3 ), the pr oposed mo v ement app lies to the last tw o types. These tw o types f orm a natural ( semantic ) class, as they can be r egar ded as generalized quantifiers o v er times/ w orlds. M o v ement of the other tw o types ar e not observ ed in Cantonese. This is because Cantonese lacks determiner ( at least in the sense of English or J apanese ) in the first p lace. Also, the pr e- v erbal negation m - is pr efixal, i.e., it is a bound morp heme (M. Y ip 1988 ), its failur e to mo v e is expected. A dditionall y , the pr oposed mo v ement ma y shift scope r elations betw een aspectual v erbs and modal v erbs on one hand and v arious quantificational elements on the other . These elements include quan- tificational subjects and topics, as w ell as quantificational adv erbials and subor dinate clauses. This diff ers fr om the cases discussed in § 4.2.3 , w hich usuall y in v ol v e onl y one type of quantificational el - ement in the discussion ( e.g., a quantificational head mo v es o v er a subject/ object quantifier or the negation ). In addition to scope eff ects with r egar d to quantificational elements, it has been argued that the pr oposed mo v ement ma y aff ect f ocus scope, or pr ecisel y , the calculation of the f ocus set. This is achiev ed b y a quantificational head escaping the scope of the f ocus operator . This type of scope eff ect has r eceiv ed little attention in the literatur e, but it nev ertheless incr eases the range of possible scope eff ects induced b y head mo v ement. F urthermor e, the pr oposed mo v ement is argued to be constrained b y an econom y condition on interpr etation, Scope Econom y ( as pr oposed in § 4.4.1.2 ), and also b y a locality condition, Shortest M o v e ( as discussed in § 4.5.2 ). This lends important support to a mo v ement anal y sis, in addition t o the eff ects of scope enrichment/ enhancement. 4.7 .2 A par all el observ ati on with p hr asal el ements W hile the discussions so far f ocus on heads that occup y the high position, a natural question is w hether the pr oposed mo v ement app lies ex clusiv el y to heads. In other w or ds, the question is w hether p hrasal 195 4.7. D ISCUSSI O NS AND IMP LI CA TI O NS elements can undergo the pr oposed scope-shifting mo v ement. In w hat f ollo w s, I argue f or a positiv e answ er with evidence fr om adv erbs that ha v e a fix ed base position. Li and Thompson ( 1981 ) observ e that adv erbs in M andarin can be divided into tw o gr oups, mov- able adv erbs and non-movable adv erbs. The descriptiv e term “( non-)mo v able” is used to indicate w hether adv erbs can appear in diff er ent positions ( such that on the surface they seem to be able to mo v e ar ound). I t is the non- mo v able ones that ar e r elev ant to the discussion her e, as non- mo v ability sug - gest that such elements ha v e a fix ed position in the sentence. This observ ation in M andarin app lies to Cantonese as w ell. F or examp le, the adv erb jau ‘ again ’ canonicall y appears after the subject. (260 ) “N on- mo v able” post -subject adv erbs a. Aaming Aaming jau again haau -dou get -able hou good singzik r esult ‘ Aaming gets good r esults again. ’ b. ?? J au again Aaming Aaming haau -dou get -able hou good singzik. r esult ‘ Aaming gets good r esults again. ’ Importantl y , jau ‘ again ’ can occup y the pr e-subject position if the subject is quantificational. N otabl y , it t ak es wide scope o v er ‘ onl y one person ’ in the deriv ed position in ( 261b ). (261) The adv erb jau ‘ again ’ and the high position a. Dak onl y jat -go one-cl jan person jau again haau -dou get -able hou good singzik. r esult ‘Onl y one person got good r esults again. ’ b. J au again dak onl y jat -go one-cl jan person haau -dou get -able hou good singzik. r esult ‘I t is again the case that onl y one person got good r esults. ’ 196 4.7. D ISCUSSI O NS AND IMP LI CA TI O NS N o te that the unacceptability of ( 260 ) is not because jau cannot modify a noun (pr o vided that dak ‘ onl y’ is v erbal b y natur e ), as sho wn in ( 262 ). (262) The adv erb jau ‘ again ’ f ollo w ed b y nominal quantifiers J au again cyunbou ev ery jan person haau -dou get -able hou good singzik r esult laa3. sfp ‘I t is again the case that ev ery one got good r esults. ’ The distribution of jau ‘ again ’ f ollo w s immediatel y fr om the pr oposed mo v ement account, and the high p osition is constrained b y Scope Econom y . T o see mor e examp les, the high position of adv erbs lik e jatzik ‘ straight/ al w a y s’ and batdyun ‘ con- tinuousl y’ is constrained in a similar w a y . The (b ) sentences in ( 263 ) and ( 264 ) sho w that the pr esence of quantificational elements (i.e., the subjects ) is crucial to the high position. (263) The adv erb jatzik ‘ straight’ and the high position a. f Z isiu at.least saam-go thr ee-cl jan/ person Aamingg Aaming jatzik straight haau -dou get -able hou good singzik. r esult ‘ A t l east thr ee peop le/ Aaming al w a y s got good r esults. ’ b. J atzik straight f zisiu at.least saam-go thr ee-cl jan/ person ??Aamingg Aaming haau -dou get -able hou good singzik. r esult ‘I t is al w a y s that at least thr ee peop le/ Aaming got good r esults. ’ (264) The adv erb batdyun ‘ continuousl y’ and the high position a. f Z isiu at.least saam-go thr ee-cl jan/ person Aamingg Aaming batd yun continuousl y haau -dou get -able hou good singzik. r esult ‘ A t l east thr ee peop le/ Aaming al w a y s got good r esults. ’ 197 4.7. D ISCUSSI O NS AND IMP LI CA TI O NS b. Batd yun continuousl y f zisiu at.least saam-go thr ee-cl jan/ person * Aamingg Aaming haau -dou get -able hou good singzik. r esult ‘I t is al w a y s that at least thr ee peop le/ Aaming got good r esults. ’ The abo v e cases of adv erbs sho w that the high position is constrained in a similar w a y to aspectual v erbs and modal v erbs. Their distribution f ollo w s fr om the pr oposed mo v ement appr oach, w hich is also constrained b y Scope Econom y . If adv erbs ar e p hrases ( as commonl y assumed), then the pr o- posed mo v ement can be generalized to app l y to both heads and p hrases. Also, Scope Econom y , as an interface condition on interpr etation, does not seem to discriminate head mo v ement fr om p hrasal mo v ement. 4.7 .3 The trigger of head mo v ement In § 4.4.3.2 , I suggested that the pr oposed HM ma y be trigger ed b y an EP P f eatur e ( without a prior Agr ee r elation ) or b y interpr etation/ scope considerations. Either w a y , the scope eff ects br ought along with this HM ar e obligatory (i.e., the lack of r econstruction eff ects ). This leads us to the pr edict that the close tie betw een HM and obligatory scope-shifting ceases to exist if the HM has a diff er ent syn- tactic trigger . This subsection discusses (i) HM trigger ed b y categorial f eatur e and (ii) HM trigger ed b y discourse f eatur e, and I sho w that the pr ediction is borne out. Let us first consider cases w her e HM is trigger ed b y categorial f eatur es. I assume with S zabolcsi ( 2010 , 2011 ) and H arizano v and Gribano v a ( 2019 ), among others, that the v erb second (V2) w or d or der in German and Dutch in v ol v es syntactic v erb mo v ement trigger ed b y categorial f eatur es ( on the C head). In German, v erb mo v ement to C does not bring along scope eff ects. ( 265a ) serv es as the baseline, w her e the v erb mo v es to C and the subject to Spec CP . The sentence unambiguousl y deliv ers the surface scope r eading. In ( 265b ), instead of the subject, the adv erbial mo v es to Spec CP . The v erb then occupies a position higher than the subject because of v erb mo v ement. Cruciall y , the sentence con v ey s the same scope r eading as in ( 265a ), suggesting that the HM of ‘begin ’ does not shift scope 198 4.7. D ISCUSSI O NS AND IMP LI CA TI O NS (i.e. it must r econstruct ). (265) (p.c. S tefan K eine ) German V2 mo v ement a. [ CP N ur onl y die the Aktienkurse stock.prices [ C’ begannen i began im in M ai M a y t i zu to steigen rise ] ] ‘In M a y , onl y stock prices begins to rise. ’ (‘ onl y’ > ‘begin ’ / *‘begin ’ > ‘ o nl y’) b. [ CP Im in M ai M a y [ C’ begannen i began nur onl y die the Aktienkurse stock.prices t i zu to steigen rise ] ] ‘In M a y , onl y stock prices begins to rise. ’ (‘ onl y’ > ‘begin ’ / *‘begin ’ > ‘ onl y’) On the other hand, head mo v ement in Dutch sho w s a slightl y diff er ent pictur e. S zabolcsi ( 2010 , 2011 ) r eports that v erb mo v ement to C in Dutch optionall y induces scope eff ects. ( 266a ) is similar to German ( 265a ), and onl y surface scope is a v ailable. H o w ev er , diff er ent fr om the German counterpart ( 265b ), Dutch ( 266b ) is ambiguous betw een a wide and lo w scope r eading of ‘begin ’ , suggesting that the v erb optionall y r econstructs. (266) (S zabolcsi 2010 , p.3 8, adapted) Dutch V2 mo v ement a. [ CP Alleen onl y M arie M ary [ C’ begon i began.3sg goede good r ollen r oles t i te to krijgen get.inf ] ] ‘Onl y M ary is such that she began to get good r oles. ’ (‘ onl y’ > ‘begin ’ / *‘begin ’ > ‘ onl y’) b. [ CP In in mei M a y [ C’ begon i began.3sg alleen onl y M arie M ary goede good r ollen r oles t i te to krijgen get.inf ] ] i. ‘Onl y M ary is such that she began to get good r oles. ’ ii. ‘I t began to be the case that onl y M ary is getting good r oles. ’ (‘ onl y’ > ‘begin ’ / ‘begin ’ > ‘ onl y’) The cases in German and Dutch illustrate that scope eff ects of HM is not obligatory if it is trigger ed 199 4.7. D ISCUSSI O NS AND IMP LI CA TI O NS b y categorial f eatur es. 54 Let us turn to cases of head mo v ement that ar e trigger ed b y discourse f eatur es. As discussed in Chapter 3, a v erb can be right dislocated to the end of the sentence in Cantonese ( with or without doubling). Arguabl y , the mo v ement is trigger ed b y some discourse f eatur e ( e.g., a def ocus f eatur e ) and the v erb lands at the CP perip hery (Lee 2017 ). The pr ediction on quantificational heads in Cantonese is that, if they mo v e f or discourse eff ects, such mo v ement need not alter scope r elations. This is borne out in ( 267 ), w her e the right dislocation of hoici ‘begin ’ giv es rise to a scopall y ambiguous sentence, suggesting that r econstruction is optional. 55 (267) Cantonese Right dislocation of quantificational heads ‘ onl y’ > ‘begin ’ / ‘begin ’ > ‘ onl y’ Dak onl y Aaming Aaming haau -dou get -able hou good singzik r esult aa3 sfp hoi ci . begin ‘Onl y Aaming begins to get good r esults. ’ This is, ho w ev er , not to sa y that discourse- f eatur e-driv en head mo v ement can nev er come with obligatory scope eff ects. Consider again the case in Shupamem (Bantu ), w her e a v erb arguabl y un- dergoes mo v ement and obligatoril y tak es wide scope o v er the subject in ( 268 b ). (268) = ( 173 ) Shupamem fr onting of aspectual v erbs (S zabolcsi 2010 , p.38) a. ‘ onl y’ > ‘begin ’ N dùù onl y M aria M aria ka past y eshe begin inget ha v e.inf ndàà good liP . r oles ‘Onl y M ary is such that she began to get good r oles’ b. ‘begin ’ > ‘ onl y’ A it -focus ka past y eshe begin ndùù onl y M aria M aria inget ha v e.inf ndàà good liP . r oles ‘I t began to be the case that onl y M ary is getting good r oles’ 54. I do not ha v e an answ er to the diff er ence betw een German and Dutch with r egar d to the r econstruction facts. 55. The pr ecise deriv ation of right dislocation should not concern us her e (f or discussions, see Chapter 3, and also L. Y .-L. Cheung ( 2009 ), Lee ( 2017 ), and Lai ( 2019 ), i.a.). The crucial observ ation her e is that the aspectual v erb finds a w a y to scope belo w dak ‘ onl y’ , w hich is una v ailable in the absence of a discourse- f eatural trigger . 200 4.8. CO N CL USI O NS W hile the natur e of the mo v ement is not exp licitl y stated in S zabolcsi ( 2010 , 2011 ), this mo v ement is accompanied with a particip le/ exp letiv e- lik e element a bef or e the v erb, a construction said to in v ol v e subject f ocus (N char e 2012 ). I t thus seems p lausible to tr eat this mo v ement as being trigger ed b y a f ocus f eatur e. S u mming up, the abo v e cases in German, Dutch, Cantonese and Shupamem suggest that, w hile head mo v ement trigger ed b y pur e EP P f eatur es/ scope considerations entails obligatory scope eff ects, other instances of head mo v ement sho w diff er ent possibilities with r egar d to scope eff ects. The r e- construction eff ects of head mo v ement ar e not unif orm ( ev en f or cases with similar syntactic triggers ) and appears to be r egulated b y some independent mechanism. I t should be r emark ed that one imp lication of the discussion in this subsection is that the empiri - cal pr operties of head mo v ement hinge on the r elev ant triggers, w hich could be a categorial f eatur e, a discourse f eatur e or pur e EP P f eatur es/ scope considerations. This patterns nicel y with p hrasal mo v e- ment, w hose pr operties of mo v ement also depend on the r elev ant triggers. W hile further comparison has to a w ait futur e r esear ch, syntactic triggers, rather than the p hrase structural status, appear to be the most r eliable basis f or the classification of mo v ement dependencies. 4.8 Conclusi ons This chapter has f ocused on the semantic eff ects of head mo v ement, w hich ar e often said to be absent fr om such mo v ement. I pr esented a no v el piece of evidence f or head mo v ement with semantic eff ects fr om Cantonese. An in-depth in v estigation into the distribution of quantificational heads such as aspectual v erbs and modal v erbs in Cantonese sho w s that these heads can occup y a non-canonical, high position in the sentence, if they ar e immediatel y f ollo w ed b y a quantificational element or a f ocused element. I pr oposed that these quantificational heads can undergo o v ert head mo v ement to a higher posi - tion and tak e scope in the landing site (i.e, scope-shifting head mo v ement ). A dditionall y , the pr oposed 201 4.8. CO N CL USI O NS mo v ement of quantificational heads is constrained b y an independentl y motiv ated condition on in- terpr etation, Scope Econom y , w hich pr ecludes semanticall y v acuous scope-shifting operations. The findings lend support to the claim that head mo v ement can induce semantic eff ects. Im- portantl y , head mo v ement can shift scope r elations, in a w a y similar to Quantifier Raising pr oposed f or nominal quantifiers. F urthermor e, Scope Econom y is sho wn to be a general constraint on both p hrasal mo v ement and head mo v ement. The conclusion of this chapter challenges the view that head mo v ement does not r esult in an y semantic eff ects - a view w hich has been used to claim that head mo v ement is non-syntactic and occurs at P F . I t is sho wn that head mo v ement is no diff er ent fr om p hrasal mo v ement in its potential to induce semantic eff ects. This motiv ates a unified theory of mo v ement of head mo v ement and p hrasal mo v ement. 202 Chapter 5 Linearizati on: doub ling eff ects of heads and p hr ases Chapter summary: This chapter investigates asymmetries in doubling among verbs, objects a nd subjects in Cantonese. I t is shown that each o f these elements has a distinct doubling pr o file in topic constructions and right dislocation: doubling is sometimes pr ohibited, r equir ed or optional. Couched in terms o f the copy theory o f movement, I suggest that that the oper ation r esponsible f or er asing copies in a movement chain is r egulated by phonological r equir ements that f ollow fr om a version o f cyclic linearization. P articularly, I pr opose that the copy-er asing oper ation can be suspended as a last r esort in cases wher e its application would otherwise violate phono- logical r equir ements imposed by cyclic linearization. The diff er ences in doubling possibility among verbs, objects and subjects f ollow fr om the availability o f the edge position o f a phase to these elements. The pr oposal derives the Cantonese doubling pattern without r ecourse to the phr ase -structur al status o f the ( non-)doubling elements and maintains that the mechanism that determines copy pr onunciation is the same f or heads and phr ases. I take this as a further piece evidence f or the unification o f head and phr asal movement, r esonating with much r ecent work on this topic. 203 5.1. INTR O D U CTI O N 5.1 Intr oducti on The goal of this chapter is tw o- f old. F irst, w hile I ha v e suggested a parallel anal y sis on v erb doubling constructions and their p hrasal counterparts in Chapter 3, a crucial diff er ence betw een them concerns the doubling eff ects: doubling eff ects in v erb doubling constructions ar e mostly obligatory , w her eas doubling eff ects of their p hrasal counterparts ar e mostly f orbidden. This chapter examines the pattern of doubling eff ects of these constructions in gr eater details and dev elop an anal y sis to account f or the asymmetries in doubling. 1 Second, with the emergence of the cop y theory of mo v ement (Chomsky 1995b , et seq. ), an inter - esting line of r esear ch has f ocusd on ho w the copies in a mo v ement chain ar e p honeticall y r ealized ( s ee Bošk o vić and N unes 2007 ; N unes 2011 , and r ef er ences ther ein ). In connection with this, the dou - bling p henomenon f ormed in v erb/ pr edicate fr onting constructions in certain languages ha v e led to v arious pr oposals w hich attempt to deriv e o v ert doubling eff ects, with r ef er ence to diff er ent com- ponents of the grammar (Landau 2006 ; Aboh and Dy ak ono v a 2009 ; T rinh 2009 ; Cheng and V icente 2013 ; H ein 2018 ). The curr ent chapt er aims to contribute to this discussion b y in v estigating the dou - bling patterns of not onl y v erbs, but also subject and objects in Cantonese, w hich each ha v e a distinct doubling pr ofile. The div erse patterns call f or an anal y sis not onl y on ho w doubling is made possible, but also, mor e importantl y , on ho w patterns of doubling ar e r egulated b y the grammar . As has been noticed in the literatur e on Chinese ( and briefl y in Chapter 3), ther e is a doubling - r elated asymmetry betw een v erb and objects in topic constructions. F or examp le, w hen the v erb soeng ‘w ant’ in ( 269 ) undergoes topicalizatio n, it is ( and must be ) doubled. 2 3 1. Specificall y , I will f ocus on t opic constructions and right dislocation, but the anal y sis can be extended to ‘ ev en ’ - f ocus constructions and copula f ocus constructions. 2. Thr oughout the paper , I consistentl y omit the tones in the r omanization of Cantonese f or simp licity , ex cept f or sentence- final particles, w her e tones ar e r epr esented to a v oid ambiguity . 3. The p henomenon is called verb doubling clefts in Cheng and V icente ( 2013 ) in M andarin, w hich in v ol v es both a topic r eading of the v erb and a v erum f ocus r eading. S ince the pr ecise discourse interpr etiv e eff ect does not bear on the arguments in the discussion in this chapter , I use the term “v erb topicalization ” as a con v enient label. See Chapter 3 f or discussions. 204 5.1. INTR O D U CTI O N (269) V erb topicalization in Cantonese (Cheng and V icente 2013 ) soeng , w ant Aaming Aaming hai co p *( soeng ) w ant sik eat jyu fish ge2 sfp ‘ As f or ( w hether he ) w ants, Aaming w ants t o eat fish (but...)’ This doubling r equir ement seems to be specifi c to v erbs. As noted in Cheng and V icente ( 2013 ) f or M a ndarin ( w hich also app lies to Cantonese ), w hen an object is topicalized, doubling is not r equir ed ( a nd i ndeed dispr ef err ed), as in ( 270 ). (270 ) Object topicalization in Cantonese ni - tiu this-cl jyu , fish, Aaming Aaming soeng w ant sik eat (?? ni - tiu this-cl jyu ) fish ‘This fish, Aaming w ants to eat. ’ The contrast betw een ( 269 ) and ( 270 ) might appear to suggest a diff er ence betw een head and p hrasal disp lacement, w her e, f or examp le, disp laced heads must be doubled w hile disp laced p hrases must not. This suggestion, ho w ev er , is empiricall y challenged b y the observ ation that a v erb can be dis- p laced to the end of the sentence with or without doubling as in ( 271 ). Associating the doubling pos- sibility with the head/ p hrase distinction thus o v ergeneralizes and w ould disallo w the non-doubling case of v erbs in right dislocation. (271) Right dislocation of v erbs in Cantonese (Lee 2017 ; Lai 2019 ) Aaming Aaming ( sik ) eat ni -di this-cl je thing aa4 q sik ? eat ‘ Aaming eats this thing?’ Another empirical challenge to such a head v s. p hrase-based appr oach is that it is possible to dou - ble a p hrase in right dislocation ( also called Dislocation Copying in L. Y .-L. Cheung 2015 ). If p hrases ar e inher entl y incompatible with doubling, right dislocation of subjects should nev er allo w tw o occur - r ences of dislocated subjects. These tw o initial observ ations suggest that an element’ s p hrasal struc - 205 5.1. INTR O D U CTI O N tural status is not sufficient to exp lain the ( non-)doubling patterns in Cantonese. (272) Right dislocation of subjects (L. Y .-L. Cheung 2009 , 2015 ) ( A aming ) Aaming soeng w ant sik eat ni -tiu this-cl jyu fish aa3 sfp A aming . Aaming ‘ Aaming w ants to eat this fish. ’ Against such a back gr ound, this chapter pursues an account on ho w and w h y doubling is pr ohib- ited , r equir ed , or optional in diff er ent cases. Specificall y , I pr opose that the operation r esponsible f or erasing copies in a mo v ement chain (i.e. Cop y Deletion ) is r egulated b y p honological r equir ements that f ollo w fr om, with some qualifications, the v ersion of p hase theory adv ocated in F o x and P esetsky ( 2005 ), namel y , C y clic Linearization ( CL ). The cor e idea will be that Cop y Deletion can be suspended as a last r esort in cases w her e its app lication w ould otherwise violate a p honological r equir ement r e- lating to C y clic Linearization. Doubling occurs as a r esult of the suspension of Cop y Deletion. The diff er ences in doubling possibility among v erbs, objects and subjects ar e deriv able fr om the a v ailabil - ity of the edge position of a p hase to such elements. The imp lications of the pr oposal ar e tw o- f old. F irst, it lends further support to C y clic Lineariza- tion, an alternativ e to Chomsky’ s v ersion of p hase theory (Chomsky 2000 , 2001 ) in the study of syn- tactic locality . Second, the pr oposal deriv es the Cantonese doubling pattern without r esour ce to the p hrase-structural status of the ( non-)doubling elements and maintains that the mechanism that de- termines cop y pr onunciation is the same f or heads and p hrases. This pr o vides a further argument f or the limited r ole of the p hrase structural status of constituents in mo v ement theories, r esonating with r ecent eff orts to unify head and p hrasal mo v ement (H artman 2011 ; F unak oshi 2012 , 2014 ; H arizano v 2019 ; H arizano v and Gribano v a 2019 ; P esetsky 2020 ). This r est of the chapter is organized as f ollo w s: § 5.2 describes the pattern of doubling in Can- tonese with r egar d to tw o constructions: topic constructions and right dislocation. § 5.3 intr oduces the framew ork and details the pr oposal. § 5.4 illustrates ho w the pr oposal deriv es the doubling asym- 206 5.2. AS YMMETRIES IN DO UB LIN G IN CANT O NESE metries in topic constructions and right dislocation. § 5.5 discusses thr ee existing accounts of dou - bling, w hich fall short of exp laining the fine-grained doubling patterns in Cantonese. § 5.6 serv es as an extension of the pr oposal and addr esses the question as to w h y v erb doubling is not al w a y s r equir ed or a llo w ed acr oss languages. § 5.7 concludes the chapter . 5.2 A s ymmetri es in doub ling in Cantonese The pattern of doubling in Cantonese is considerabl y intricate. In w hat f ollo w s, I discuss the pattern of subjects, v erbs and objects in topic constructions and right dislocation, w her e each of the f ormer elements has its o wn doubling pr ofile. As will be seen, contrasts betw een the doubling possibilities in topicalization and right dislocation suggest that the dir ectionality of disp lacement p la y s an important r ole in determining w hether doubling is possible or not. A first asymmetry in doubling has alr eady been noted in the intr oduction. The r elev ant examp les ( 269 ) and ( 270 ) ar e r epeated belo w in ( 273 ). The crucial observ ation is that, w hile both v erbs and objects can be topicalized, v erbs must be doubled 4 , but objects cannot be dou bled. (273) T opic constructions a. v erbs, =( 269 ) Soeng , w ant Aaming Aaming hai co p *( soeng ) w ant sik eat jyu fish ge2. sfp ‘ As f or ( w hether he ) w ants, Aaming w ants to eat fish (bu t...)’ b. objects, =( 270 ) N i - tiu this-cl jyu , fish, Aaming Aaming soeng w ant sik eat (?? ni - tiu this-cl jyu ). fish ‘This fish, Aaming w ants to eat. ’ 4. I t is also possible to topicalize the lo w er v erb sik ‘ eat’ , w her e doubling is obligatory . (i) S ik , eat Aaming Aaming hai foc soeng w ant *( sik ) eat jyu fish ge2. sfp ‘ As f or ( w hether he w ants to ) eat, Aaming w ants to eat fish (but...)’ 207 5.2. AS YMMETRIES IN DO UB LIN G IN CANT O NESE T w o r emarks ar e in or der . F irst, v erb topicalization is diff er ent fr om object topicalization in that w hen a v erb is topicalized, the pr esence of the copula hai is str ongl y pr ef err ed. N o such pr ef er ence is observ ed in object topicalization. W hile it is clear that the copula contributes to a v erum f ocus r eading (Cheng and V icente 2013 ), it is less clear w h y it is associated with v erb topicalization, but not object topicalization. I will not pursue an exp lanation of this diff er ence, ho w ev er . Second, ther e is a diff er ence in terms of acceptability with r egar d to the absence of doubling in v erb topicalization (i.e. ( 273a ) without the second occurr ence of soeng ‘w ant’) and the pr esence of doubling in object topicalization (i.e. ( 273b ) with the second occurr ence of ni- tiu jyu ‘this fish ’). W hile both ar e judged as deviant, the latter is judged as r edundant and is slightl y mor e acceptable than the f ormer . 5 I mark sentences with a r eported sense of r edundancy with ??, instead of *, to indicate the diff er ence in acceptability . The diff er ence seems to suggest a violation of diff er ent grammatical princip les. I will r eturn to this point in § 5.4.3 . T o see a second asymmetry of a similar kind, let us turn to right dislocation. I t has been r eported that both v erbs and objects can be dislocated to the right of sentence- final particles (Lee 2017 ; Lai 2019 ). H o w ev er , w e observ e that v erbs ar e optionally doubled, w her eas objects can har dl y be doubled ( as it giv es rise to a hea vy sense of r edundancy ). (274) Right dislocation a. v erbs, =( 271 ) Aaming Aaming ( sik ) eat ni -di this-cl je thing aa4 q sik ? eat ‘ Aaming eats this thing?’ b. objects Aaming Aaming sik eat (?? ni -di this-cl je ) thing aa4 q ni -di this-cl je ? thing ‘ Aaming EA TS this thing?’ N ote that f or ( 274a ), the sentences with and without a doubled v erb sho w diff er ent f ocus inter - 5. I thank A udr ey Li f or pointing out this diff er ence to me. 208 5.2. AS YMMETRIES IN DO UB LIN G IN CANT O NESE pr etations. W hen the v erb is doubled, the v erb r eceiv es a f ocus interpr etation. In contrast, w hen the v erb is right dislocated without doubling, the object r eceiv es a f ocus interpr etation. I will r eturn to this observ ation in § 5.4.2 . F inall y , let us tak e the pattern of subjects into consideration, w her e w e observ e y et another asym- metry: topicalized subjects do not go w ell with doubling ( as it similarl y giv es rise to an air of r edun- dancy ), w her eas right -dislocated subjects ar e optionally doubled. N ote that doubling of the subjects in ( 275b ) is not judged as r edundant, unlik e the doubling of objects in ( 273b ) and ( 274b ). Again, ther e is an interpr etiv e diff er ence associated with doubling: the subject is f ocused w hen doubled. W hen the subject is not doubled, the w hole v erb p hrase is f ocused. (275) a. T opic constructions subjects A aming Aaming ( n e ), to p (?? A aming ) Aaming soeng w ant sik eat ni -tiu this-cl jyu. fish ‘ As f or Aaming, (he ) w ants to eat this fish. ’ b. Right dislocation subjects, =( 272 ) ( A aming ) Aaming soeng w ant sik eat ni -tiu this-cl jyu fish aa3 sfp A aming . Aaming ‘ Aaming w ants to eat this fish. ’ T able 5.1 belo w pr o vides a summary of the doubling pr ofile of v erbs, subjects and objects. The patterns can be described as f ollo w s: (i) object doubling is generall y pr ohibited (in both topic con- structions and right dislocation ); (ii) verb doubling is obligatory in topic constructions, but optional in right dislocation; (iii) subject doubling is pr ohibited in topic constructions, but optional in right dislocation. S ubject V erb Object T opic constructions pr ohibited obligatory pr ohibited Right dislocation optional optional pr ohibited T able 5.1 : Doubling asymmetries in Cantonese 209 5.3. P R O POSAL: CY CLI C LINEARIZA TI O N AND CO P Y D ELETI O N SUSP ENSI O N Capturing such div erse patterns of doubling in Cantonese in a non-trivial w a y pr esents a sig - nificant challenge. One thing that seems clear , ho w ev er , is that the possibility f or an element to be doubled does not immediatel y f ollo w fr om an y head/ p hrase distinction, because it is not the case that heads al w a y s r equir e doubling ( e.g. ( 274a )) or that p hrases can nev er b e doubled ( e.g. ( 275b )). 5.3 Pr oposal: C y cli c Linearizati on and Cop y Del eti on suspen - si on U nder the cop y theory of mo v ement, w hile it is generall y agr eed that some operation is r esponsible f or deleting r edundant copies in the pr ocess of linearization ( e.g. Cop y Deletion, see Chomsky 1995b ; N unes 1995 , 2004 ), opinions v ary as to w hat in the grammar allo w s or ev en r equir es the surviv al of a second cop y . I pr opose that the deletion of a (lo w er ) cop y ma y be suspended if it violates linearization r equir ements imposed b y C y clic Linearization (F o x and P esetsky 2005 ). I first o v erview the pr oposal of C y clic Linearization, CL, and then go into the details of m y pr oposal. 5.3.1 C y cli c Linearizati on F o x and P esetsky ( 2005 ) pr opose that syntactic structur e is linearized cy clicall y . P articularl y , in each domain w her e all ( necessary ) syntactic operations ar e app lied and the structur e is r eady to be lin- earized ( e.g. v P and CP ), it is Spelled-Out . U pon Spell -Out, Or dering S tatements ( OS ), the or dering inf ormation among o v ert elements, is established. Cruciall y , OS must be pr eserv ed b y o v ert elements in the final output. OS can thus be consider ed as p honological r equir ements deriv ed along the syn- tactic deriv ation. Also, OS ar e cumulativ e and cannot be o v erwritten. I imp lement the idea of CL under the cop y theory of mo v ement. 6 I suggest that tw o operations tak e p lace at each instance of Spell -Out: 6. In F o x and P esetsky ( 2005 ), mo v ement is construed as an operation of r emerge , w hich establishes multi -dominance r elations among the elements. 210 5.3. P R O POSAL: CY CLI C LINEARIZA TI O N AND CO P Y D ELETI O N SUSP ENSI O N (276) A cop y -theor etic imp lementation of CL A t each Spell -Out domain, tw o independent operations app l y one after the other: (i) Cop y Del eti on ( CD , typicall y deleting the lo w copies ), f ollo w ed b y (ii) Linearizati on (LIN , establishing Or dering S tatements ). Let us consider the tw o scenarios in ( 277 ) and ( 278 ) belo w . In both scenarios, LIN occurs at domain D , establishing the OS D : X < Y < Z . In the next domain D’ , some element is merged. The tw o scenarios div erge fr om he r e. In Scenario 1, mo v ement of X to D’ and the deletion of its lo w er cop y ( mark ed in gra y color ) w ould not violate the pr eviousl y established OS, i.e. X still pr ecedes both Y and Z. H o w ev er , in Scenario 2, mo v ement of Y to D’ poses a linearization pr oblem. This is because w hen CD app lies to the lo w cop y of Y at the Spell -Out of D’ , the r equir ement that X pr ecedes Y is no longer obey ed. (277) Scenario 1 (LIN D ! M o v e X ! CD! LIN D’ ) OS D’ : X < < D (X < Y < Z) [ D’ ... X [ D X Y Z ]] (278) Scenario 2 (LIN D ! M o v e Y ! CD! LIN D’ ) *OS D’ : Y < < D (X < Y < Z) * [ D’ ... Y [ D X Y Z ]] Importantl y , it is not the case that mo v ement of non-edge elements is nev er possible. I t is possible if a non-edge element mo v es successiv e cy clicall y out of a Spell -Out domain. F or examp le, if Y mo v es to the edge of D bef or e it mo v es out to D’ , then the OS established at D w ould be diff er ent fr om Sce- nario 1 and 2: Y < X < Z . S ubsequent mo v ement of Y to a higher domain as depicted in Scenario 3 is possible, as app l ying CD to the lo w copies of Y w ould not violate an y OS, i.e. Y still pr ecedes X and Z w hen Spelled-Out at D’ . 211 5.3. P R O POSAL: CY CLI C LINEARIZA TI O N AND CO P Y D ELETI O N SUSP ENSI O N (279) Scenario 3 ( M o v e Y within D! CD! LIN D ! M o v e Y ! CD! LIN D’ ) OS D’ : Y < < D (Y < X < Z) [ D’ ... Y [ D Y X Y Z ]] Diff ering fr om standar d assumptions about p hases (particularl y the v ersion in Chomsky 2000 , 2001 ), CL opens certain other possibilities f or mo v ement of non-edge elements. In Scenario 4, non- edge elements can mo v e acr oss edge elements if the mo v ement of the f ormer is f ollo w ed b y some “ compensating mo v ement” of the latter that pr eserv es the or dering r elations. F or examp le, the mo v e- ment of Y is allo w ed if X also mo v es to a position higher than Y . Consequentl y , the or der betw een X and Y is pr eserv ed and ther e is no violation of an y OS. 7 (280 ) Scenario 4 (LIN D ! M o v e Y & M o v e X ! CD! LIN D’ ) OS D’ : X < Y < < D (X < Y < Z) [ D’ ... X ... Y [ D X Y Z ]] 5.3.2 Cop y Del eti on suspensi on Against this back gr ound, I pr opose that the app lication of Cop y Deletion is constrained b y lineariza- tion r equir ements imposed b y CL. (281) Cop y Deletion suspension Cop y Deletion is suspended as a last r esort if its app lication violates linearization r equir ements imposed b y CL. Cruciall y , the doubling p henomenon arises as a r esult of the suspension of CD . R ecall that w hen a non- edge element exits a domain D to another domain D’ , the OS established at D is violated. H o w ev er , the violation is a v oided if the non-edge element is pr onounced (i.e. doubled). Schematicall y , consider Scenario 5 belo w: 7. This is arguabl y the case f or object shift in Scandina vian languages, see F o x and P esetsky ( 2005 ) f or extensiv e dis- cussion. 212 5.3. P R O POSAL: CY CLI C LINEARIZA TI O N AND CO P Y D ELETI O N SUSP ENSI O N (282) Scenario 5 (LIN D ! M o v e non-edge ! CD suspensi on! LIN) OS D’ : Y < < D (X < Y <Z) [ D’ ... Y [ D X Y Z ]] W hen D i s Spelled-Out, the OS: X > Y is satisfied b y the pr onunciation of the lo w er cop y of Y . A t the later Spell -Out of D’ , the OS: Y > X is also satisfied b y virtue of the higher cop y of Y . As such, the mo v ement of Y does not violate an y OS. A potential concern, ho w ev er , is that the tw o OS abo v e r equir e that Y must pr ecede X and X must pr ecede Y , and cruciall y , b y transitivity , Y must pr ecede Y . U nder a multi -dominance or r emerge ap- pr oach to mo v ement as originall y assumed in F o x and P esetsky ( 2005 ), this r equir ement constitutes a linearization contradiction since a pr ecedence r elation cannot be r eflexiv e, i.e. ther e is no w a y f or Y to pr ecede or to be pr eceded b y itself. W ith the adaptation of CL to the cop y theory of mo v ement, the conclusion that Y must pr ecede Y need not be a contradiction. W hile the tw o copies ar e identical to each other (in terms of f eatural mak eup ), they ar e tw o separate elements in a chain of mo v ement. A pr ecedence r elation betw een tw o copies should ther ef or e, in princip le, be possible. The question is that ho w the computational sy stem diff er entiates the tw o copies if they ar e indeed identical. One w a y is to intr oduce indices to copies, but this w ould violate the Inclusiv eness Condition (Chomsky 1995b ); another is suggested in N unes ( 2004 , p.165 fn. 15): if a new term is intr oduced into the computation without r educing the numeration, the computational sy stem “kno w s” that a cop y of some syntactic object has been cr eated. The curr ent pr oposal does not hinge on w hich choice is adopted her e. R ele- v ant to us is that the r equir ement “ Y must pr ecede Y” can be satisfied, pr o vided that the computational sy stem is capable of diff er entiating copies and establishing a pr ecedence r elation among them. Bef or e discussing the consequences of the pr oposal, I will mak e a f ew assumptions in the upcom- ing discussion. F irst, I assume that right dislocation in Cantonese in v ol v es syntactic mo v ement within a m ono-clausal structur e, a position def ended in L. Y .-L. Cheung ( 2009 ), Lee ( 2017 ), and Lai ( 2019 ). Second, f ollo wing Cheng and V icente ( 2013 ), Lee ( 2017 ), and Lai ( 2019 ), I assume that v erb mo v e- ments ar e in v ol v ed in both v erb topicalization and the right dislocation of v erbs, i.e. the disp laced 213 5.3. P R O POSAL: CY CLI C LINEARIZA TI O N AND CO P Y D ELETI O N SUSP ENSI O N v erbs ar e not base generated. 8 The arguments f or a mo v ement anal y sis comes fr om the observ ation that the disp laced v erbs sho w connectivity eff ects with the other cop y . These include island eff ects and lexical identity eff ects. The f ormer is illustrated with the sentences in ( 283 ), w her e the doubled v erbs cannot be separated b y an island boundary . (283) a. V erb topicalization (based on Cheng and V icente ( 2013 , p.8)) * T ai, see ngo I tongji agr ee [ NP go-go that -cl k eoi s/he hai co p tai -gw o see-exp ge ge jigin ], opinion batgw o... but ‘ As f or seeing, I agr ee with the opinion that s/he has indeed seen it, but... ’ b. Right dislocation of v erbs (based on Lee ( 2017 , p.65)) * Aaming Aaming zipsau - m-dou accept - not -able [ NP Aafan Aafan ( sik ) kno w gong speak sap- zung ten-cl jyujin language ge ge sisat ] fact aa3 sfp sik . kno w ‘ Aaming cannot accept the fact that Aafan can speak ten languages. ’ A dditionall y , the tw o v erbs must be lexicall y identical to each other , an observ ation tak en to be evidence f or mo v ement (Cable 2004 ; V icente 2007 ; Cheng and V icente 2013 ). I illustrate the idea with the v erb caa ‘ check’ and cek ‘ check’ ( an English loan w or d). The identity eff ect is observ ed in spite of the semantic identity of the tw o v erbs. (284) a. V erb topicalization f Caa /* Cekg , check/ c heck ngo I hai co p caa -gw o check -exp ni -go this-cl jan, person batgw o... but As f or checking, I ha v e check ed this person, but... ’ 8. See Chapter 3 f or a detailed pr oposal of a head mo v ement anal y sis on these constructions. 214 5.4. D ERIVIN G THE AS YMMETRIES IN DO UB LIN G b. Right dislocation of v erbs N ei y ou soeng w ant cek check ni -go this-cl jan person aa4 q f cek /* caa ?g check/ check ‘Do y ou w ant CHECK this person?’ Lastl y , I assume that a head cannot mo v e into its o wn specifier . This assumption ma y f ollo w fr om a v ersion of anti - locality constraint w hich pr ohibits mo v ement operations that ar e too “local” (Abels 2003 , i.a.); or it ma y be due to the lack of motiv ation: a head need not mo v e to its o wn specifier to check f eatur es (if ther e is an y ). Either assumption w ould rule out the mo v ement step illustrated belo w , using the v head as an examp le. (285) Illicit v - mo v ement v P v v P subject v ’ v VP 5.4 Deri ving the as ymmetri es in doub ling N o w , w e ha v e all the ingr edients w e need to account f or the doubling asymmetries in Cantonese. § 5.4.1 and § 5.4.2 illustrate ho w the pr oposal deriv es doubling asymmetries in topics const ructions and right dislocation, r espectiv el y . § 5.4.3 r eturns to the acceptability issue mentioned in passing in § 5.2 . § 5.4.4 discusses one mor e asymmetry w hich also falls out fr om the curr ent pr oposal. 215 5.4. D ERIVIN G THE AS YMMETRIES IN DO UB LIN G 5.4.1 Li cit and illi cit cases in topi c constructi ons I start with the doubling case in v erb topicalization, f ollo w ed b y the obligatory absence of doubling f or subjects and objects. R ecall that v erb doubling in topic constructions is obligatory . The r elev ant examp le is r epeated belo w as ( 286 ). (286) V erb topicalization =( 269 ) Soeng , w ant Aaming Aaming hai co p soeng w ant sik eat jyu fish ge2. sfp ‘ As f or ( w hether he ) w ants, Aaming w ants to eat fish (but...)’ U nder the curr ent pr oposal, ( 286 ) has the deriv ation giv en in ( 287 ). F irst, in ( 287 a ), the v P headed b y soeng ‘w ant’ is built, with Aaming being the subject and sik jyu ‘ eat fish ’ the ( clausal) comp lement. 9 The copula hai is assumed to occup y a v P - internal position sandwiched betw een the main v erb and the subject. N ote that soeng does not ( and cannot ) mo v e to the edge of v P . In ( 287 b ), the v P is Spelled- Out, and the or der betw een the subject/ hai and the v erb soeng ‘w ant’ is fix ed. ( 287 c ) indicates the mo v ement of soeng to a T opic position, one that is higher than the subject. 10 In ( 287 b ), at the final Spell -Out domain T opic P , cruciall y , CD of soeng ‘w ant’ is suspended, because deleting its lo w er cop y w ould violate the OS established in v P . The w hole structur e is thus linearized with the pr esence of tw o copies of soeng . 11 12 9. The deriv ation is also compatible with an anal y sis w her e soeng ‘w ant’ is base generated at V and then undergoes head mo v ement to v (H uang 1994 , 1997 ; T ang 1998b ), but this step does not bear on the pr oposal. 10. I abstract o v er the standar d subject mo v ement to Spec TP f or its irr elev ance. 11. F or simp licity , the sentence- final particle ge2 w hich is external to the v P is not sho wn in the der iv ation. 12. An anon ymous JEAL r eview er points out that modal v erbs such as wui ‘will’ can be doubled as w ell, as in (i). (i) K eoi s/he wui will lai come hoiwui join.meeting gaa3 sfp wui . will ‘S/he will come to the meeting. ’ One potential concern is that modal v erbs might occup y a position bey ond v P such that their r elativ e position with the v P - internal elements is not fix ed upon the Spell -Out of v P and hence doubling is expected not to be necessitated. I t should be noted, ho w ev er , that the modal v erbs ha v e been argued to be lexical pr edicates, heading a V / v position (Lin and T ang 1995 ). In such case, modal v erbs double in the same w a y as soeng ‘w ant’ . Alternativ el y , it is possible that modal v erbs ar e also p hase heads. If w e f ollo w a contextual appr oach to p hasehood as adv ocated b y Bošk o vić ( 2014 ) w her e the highest position of an extended pr ojection constitutes a p hase. In such case, doubling is as expected. 216 5.4. D ERIVIN G THE AS YMMETRIES IN DO UB LIN G (287) Deriv ation of ( 286 ) a. Building of v P (headed b y soeng ‘w ant’) [ v P Aaming hai soeng sik jyu ] b. Spell -Out of v P ( CD does not app l y or app lies v acuousl y ) ( CD )! LIN v P ; OS v P : A aming < hai < soeng < sik < jyu c. V erb mo v ement f or topicalization [ T opic P soeng ... [ v P Aaming hai soeng sik jyu ]] d. Spell -Out of T opic P CD s uspension! LIN T op P ; OS T op P : soeng < A aming < hai < soeng < sik < jyu On the other hand, objects beha v e diff er ent fr om v erbs in topic constructions. Doubling of objects is disallo w ed. R ecall ( 270 ), r epeated belo w as ( 288 ). (288) Object topicalization = ( 270 ) N i - tiu this-cl jyu , fish, Aaming Aaming soeng w ant sik eat (?? ni - tiu this-cl jyu ). fish ‘This fish, Aaming w ants to eat. ’ I suggest that the crucial diff er ence betw een v erbs and objects does not lie in their p hrase structural status, but in their possibility of mo v ement to Spec v P . W hile a v erb cannot mo v e to its o wn specifier , the Spec position is a v ailable to objects. 13 In cases w her e the object mo v es to Spec v P , the OS at the Spell -Out of v P becomes: O < S < V . M o v ement of the object to a higher domain w ould not suspend CD at Spell -Out. Lo w er copies ar e deleted b y CD , r esulting in the absence of doubling ( cf. Scenario 13. This diff er ence betw een head mo v ement and p hrasal mo v ement is not due to some inher ent pr operties of head mo v ement; rather , as suggested in the discussion ar ound ( 285 ), this might be attributed to a general anti - locality constraint on mo v ement or to the lack of motiv ation. 217 5.4. D ERIVIN G THE AS YMMETRIES IN DO UB LIN G 3). The deriv ation is schematicall y r epr esented in ( 289 ). (289) The schematic deriv ation of object topicalization Object mo v ement! CD! LIN T op P ; OS T op P : O < S < V [ T op P O ... [ v P O S V O ]] The same line of r easoning app lies to subjects in topic constructions w her e doubling is not al - lo w ed. The onl y diff er ence is that subjects do not mo v e into Spec v P; instead, they ar e base generated ther e. Independentl y of their deriv ational histories, mo v ement of neither subjects nor objects w ould trigger suspension of CD . (290 ) The schematic deriv ation of subject topicalization S ubject mo v ement! CD! LIN T op P ; OS T op P : S < V < O [ T op P S ... [ v P S V O ]] T w o r emarks ar e in or der . F irst, w hile the a v ailability of a successiv e cy clic mo v ement path of objects r enders doubling unnecessary , one ma y w onder w h y doubling is disallo w ed. Concr etel y , if the object does not stop at Spec v P , it w ould be linearized to the right of the subject and the v erb at the Spell -Out of v P: S < V < O . W hen the object subsequentl y mo v es out of v P , this should r esult in CD suspension in the same w a y as v erb topicalization. I suggest that doubling of objects is disallo w ed because CD can onl y be suspended as a last r esort . If successiv e cy clic mo v ement is a v ailable to objects, it must app l y (hence no suspension of CD ). The last r esort natur e of CD suspension is pr obabl y r elated to an econom y princip le that pr ef ers a structur e with minimal number of copies, w hich in turn pr ef ers the app lication of CD w her ev er possible. In sum, under the curr ent pr oposal, the asymmetry betw een v erbs on one hand and objects and subjects on the other in topic constructions is deriv able fr om the structural position (i.e. the launch- ing site ). I t hinges on the possibility to occup y Spec v P . V erbs ar e “ special” not because they ar e heads, but because they fail to mo v e to Spec v P . In contrast, the Spec v P position is a v ailable to objects and subjects, h ence t he abse nce of doubli ng. The appr oach to the doubling p henomenon mak es no r ef er - 218 5.4. D ERIVIN G THE AS YMMETRIES IN DO UB LIN G ence to the head-p hrase distinction, w hich in turn a v oids the o v ergeneralization pr oblem mentioned in the intr oduction. 5.4.2 Li cit, illi cit and opti onal cases in right disl ocati on 5.4.2.1 Li cit cases The situations in right dislocation ar e mor e comp licated since doubling is optional in some cases. Let us start with the licit and illicit cases. The line of r easoning will be largel y similar to w hat w e ha v e seen f or topic constructions, with the onl y diff er ence being the dir ection of mo v ement. W e ha v e seen that doubling of a leftwar d-moving v erb is licit because it is pr eceded b y some v P - internal elements. W e no w also ha v e the opposite case, w her e a rightwar d-moving v erb is doubled because it is f ollowed b y some v P - internal elements. Doubling of mo ving v erbs ar e possible in the pr esence of objects or embedded v erbs ( as in ( 291 a ) and ( 292 a ), r espectiv el y ), w hich is otherwise degraded ( as in ( 291 b ) and ( 292 b )). (291) V erb doubling allo w ed in the pr esence of objects a. K eoi he sik eat ni -di this-cl je thing aa4 Q sik ? eat ‘ H e EA TS this thing?’ b. ?? K eoi he sik eat aa4 Q sik ? eat Intended: ‘H e EA TS?’ (292) V erb doubling allo w ed in the pr esence of embedded v erbs a. K eoi he soeng w ant heoi go gaa3 sfp soeng . w ant ‘H e W ANTS to go. ’ b. ?? K eoi he soeng w ant heoi go gaa3 sfp heoi . go ‘H e w ants to GO . ’ The contrast f ollo w s fr om the curr ent pr oposal: w hen the v P headed b y the mo ving v erb is Spelled- Out, the or der of the v erb with r egar d to the object and the embedded v erb is fix ed. The v erb must 219 5.4. D ERIVIN G THE AS YMMETRIES IN DO UB LIN G pr ecede them in the final w or d or der . A ccor dingl y , at the final Spell -Out, CD on the lo w er cop y of the mo ving v erb has to be suspended (hence doubling occurs ), or it w ould violate the OS. The deriv a- tions of these cases ar e schematicall y r epr esented in ( 293 ) and ( 294 ), r espectiv el y . In w hat f ollo w s, f or simp licity , I onl y detail the operations within v P (if an y ) and the OS established at the Spell -Out of v P . S ince OS cannot be o v erwritten, the final w or d or der must obey this OS. 14 15 (293) The schematic deriv ation of ( 291 ) OS v P : S <V < O [ v P S V O ] sfp V (294) The schematic deriv ation of ( 292 ) [ v P S V1 [ TP ... V2 ... ]] sfpf V 1/??V2g OS v P : S <V1 <V2 In a similar v ein, subjects ar e doubled in right dislocation f or the same r eason that v erbs ar e dou - bled in right dislocation: subjects ar e at least f ollo w ed b y a v erb ( and potentiall y also an object ). This giv es rise to the OS v P w hich dictates S < V < O . Right dislocating the subject to the right of the v erb triggers CD suspension at a later point of Spell -Out, r esulting in doubling. The r elev ant examp le and its deriv ation ar e giv en belo w . (295) S ubject doubling in right dislocation A aming Aaming soeng w ant sik eat ni -tiu this-cl jyu fish aa3 sfp A aming . Aaming ‘ AAMIN G w ants to eat this fish. ’ 14. F or illustrativ e purposes, I assume a rightw ar d mo v ement appr oach of right dislocation, but the anal y sis is compat - ible with w hatev er mechanism that r enders the v erb ending up in the rightmost position ( e.g. a Ka ynean-sty le multip le leftw ar d mo v ement ). 15. The pr ecise position of the sentence- final particles is immaterial her e, as long as they occup y some position in the CP perip hery (L. L.-S. Cheng 1991 , i.a.). I t ma y be head- initial (p lus TP mo v ement ) or head- final. F or discussions on this issue, see L. Y .-L. Cheung ( 2009 ). 220 5.4. D ERIVIN G THE AS YMMETRIES IN DO UB LIN G (296) The schematic deriv ation of ( 295 ) OS v P : S <V < O [ v P S V O ] sfp S I t is notew orth y that the right dislocated elements ma y contain both a subject and an adv erb, as in ( 297 ). S ince the subject and the adv erb do not f orm one constituent, ( 297 ) ma y in v ol v e multip le operations of right dislocation, i.e. they ar e right dislocated separatel y , illustrated in ( 298 ). (297) N go I jau again sihaa try sin1 sfp ngo I jau . again ‘Let me try as w ell. ’ (Lai 2019 , p.254) (298) The schematic deriv ation of ( 297 ) OS v P : S < again <V [ v P S again V ] sfp S again Importantl y , the r elativ e or der betw een the subject and the adv erb ar e pr eserv ed w hen right dislo- cated. This f ollo w s naturall y fr om the curr ent anal y sis since their or der is fix ed w hen the v P is Spelled- Out. v 5.4.2.2 Illi cit cases T urning to the illicit doubling cases specific to objects, the una v ailability of object doubling f ollo w s fr om the fact that an object is typicall y at the right edge of a v P . Right dislocation of the object after the Spell -Out of v P w ould be subject to CD . (299) a. N o doubling in right dislocation of objects =( 274b ) Aaming Aaming sik eat (?? ni -di this-cl je ) thing aa4 q ni -di this-cl je ? thing ‘ Aaming EA TS this thing?’ b. The schematic structur e of ( 299a ) 221 5.4. D ERIVIN G THE AS YMMETRIES IN DO UB LIN G OS v P : S < V < O [ v P S V O ] sfp O A straightf orw ar d pr ediction is that if ther e ar e elements that f ollo w objects, the doubling of ob- jects should be possib le. This is indeed the case. Consider the f ollo wing examp les with a duration p hrase and an indir ect object (in a ditransitiv e structur e ), r espectiv el y: (300 ) a. Duration p hrases Aaming Aaming tai - zo r ead-perf ni -bun this-cl s yu book saam-go thr ee-cl zong hour laa3 sfp ni -bun this-cl s yu . book ‘ Aaming has r ead T HIS BOO K f or thr ee hours. ’ b. Indir ect object Aaming Aaming bei - zo giv e-perf ni -bun this-cl s yu book Aafan Aafan laa3 sfp ni -bun this-cl s yu . book ‘ Aaming has gi v en THIS BOO K to Aafan. ’ Giv en sufficient contexts ( e.g. one in w hich the speak er is contrasting the book under discussion with other books ), both sentences ar e acceptable, at least disp la ying a contrast with ( 299a ). Assuming that both the duration p hrase and the indir ect object ar e within the v P , the ir r elativ e w or d or der is fix ed w hen v P is Spelled-Out. Doubling is possible f or the object since it is no longer at the right edge of v P , as illustrated belo w . (301) The schematic structur e of sentences in ( 300 ) OS v P : S < V < O < 3- hours/ Aafan [ v P S V O 3- hours/ Aafan ] sfp O I t is instructiv e to note that Lai ( 2019 ) argues that object doubling ( e.g. the case in ( 299a )) cannot be ruled out b y a pur e p honological consideration, such as the a v oidance of p honological identity . H e supports this claim with the f ollo wing examp le. In ( 302 ), the object keoi ‘ s/he’ is p honologicall y identical to the right dislocated element ( w hich is co- index ed with the subject ). Y et, the sentence is 222 5.4. D ERIVIN G THE AS YMMETRIES IN DO UB LIN G acceptable. 16 (302) K eoi i s/he zungji lik e k eoi j her /him aa3 sfp k eoi i . s/he ‘S/he lik es her /him. ’ (Lai 2019 , p.246, with adaptations ) A c cor dingl y , the sentence in ( 299a ) with a doubled object is not ruled out b y p honological identity but b y an econom y consideration in w hich lo w er copies ha v e to be deleted in general. The same can be said f or the tw o sentences in ( 291 b ) and ( 292 b ). 17 5.4.2.3 Opti onal cases The r emaining question is w h y ther e ar e cases of optional doubling. M or e specificall y , ho w can v erbs and subjects mo v e without doubling (in a w a y similar to right dislocating objects )? I pr opose that this is made possible b y independent mo v ement operations that r e-arrange the elements in the v P bef or e Spell -Out, as will be described belo w . As a r esult of these mo v ements, v erbs and subjects ma y appear on the right edge upon Spell -Out of v P and ma y mo v e without doubling, lik e the case of objects. Let us start with the case of non-doubling v erbs. In examp les lik e ( 303 a ), I suggest that ther e is object movement bef or e the Spell -Out of v P , establishing a diff er ent OS v P : S < O < V . 18 S ubsequent right dislocation of the v erb is f ollo w ed b y CD , w hich is not suspended, r esulting in no doubling. 16. Inter estingl y enough, Lai ( 2019 ) observ es that co- indexation betw een the object and the right dislocated element is disallo w ed, w hich is in line with the observ ation that objects cannot be doubled. 17. An anon ymous JEAL r eview er raises concerns o v er cases lik e (i), w her e the subject and the dislocated element ar e co- index ed, but they ar e not identical ( r ef err ed to as I mperf ect Copying in L. Y .-L. Cheung ( 2015 )). (i) A aming i Aaming soeng w ant sik eat ni -tiu this- cl jyu fish aa3 sfp k eoi i . s/he ‘ Aaming w ants to eat this fish ’ Right dislocation in v ol ving Imperf ect Cop ying poses a general challenge to existing mo v ement appr oaches ( e.g. L. Y .-L. Cheung 2009 ; Lee 2017 ; Lai 2019 ). The acceptability of these cases is sometimes tak en to motiv ate a bi -clausaal anal y sis of right dislocation, as pursued in L. Y .-L. Cheung ( 2015 ). This imp lies that the right dislocated elements in Cantonese ma y ha v e diff er ent deriv ational possibilities. Indeed, ev en within the mo v ement appr oaches, the status of the dislocated element is not the same: it ma y be in its base generation position (L. Y .-L. Cheung 2009 ) or in a deriv ed position (Lee 2017 ). 18. I assume that the object mo v ement is achiev ed b y some ‘tucking - in ’ operation, landing in a position belo w the subject (N . Richar ds 2001 ). 223 5.4. D ERIVIN G THE AS YMMETRIES IN DO UB LIN G (303) a. N o doubling in right dislocation of v erbs K e oi he ni -di this-cl je thing aa4 Q sik i ? eat ‘H e eats THIS THIN G?’ b. The schematic deriv ation of ( 303 a ) O bject mo v ement! CD! LIN v P ; OS v P : S < O < V [ v P S O V O ] sfp V Ther e is independent evidence f or the pr oposed object mo v ement. F irst, it is not parasitic on right dislocation and can be app lied independentl y . (304) K eoi he [ ni -di this-cl je ] i thing sik eat t i gaa4? q ‘H e eats THIS THIN G?’ Importantl y , if the nominal in the object position cannot undergo object mo v ement ( e.g. bar e noun indefinites ) as in ( 305 a ), the v erb cannot be right dislocated either as in ( 305 b ). This suggests that right dislocation of v erbs counts on the successful app lication of object mo v ement. (305) a. * Aaming Aaming je i thing sik eat t i aa4? q Int.: ‘ Aaming eats?’ b. * Aaming Aaming je thing t i aa4 q sik i ? eat Int.: ‘ Aaming eats?’ F urther support the the corr elation betw een successful object mo v ement and right dislocation v erb without doubling comes fr om ditransitiv e structur es. Observ e that w hile right dislocation of the v erb bei ‘ giv e’ is allo w ed in ( 306 ), it must be doubl ed. 19 19. I thank an anon ymous JEAL r eview er f or pointing out this contrast. 224 5.4. D ERIVIN G THE AS YMMETRIES IN DO UB LIN G (306) Aaming Aaming *( bei ) giv e [ ni -bun this-cl syu ] DO book ngo I O I aa3 sfp bei . giv e ‘ Aaming giv es this book to me. ’ The doubling case f ollo w s straightf orw ar dl y fr om the pr oposal ( see ( 300 ) abo v e ). The question is w h y v erb doubling is obligatory . I suggest that doubling is f or ced because the indir ect object ngo fails to undergo object mo v ement f or independent r easons. 20 I ts immobility is evidenced b y the f ollo wing paradigm, independentl y of right dislocation of v erbs. (307) a. (baseline ) Aaming Aaming bei giv e [ ni -bun this-cl syu ] DO book ngo I O I aa3. sfp ‘ Aaming giv es this book to me. ’ b. Aaming [ ni -bun s yu ] DO bei t DO ngo I O aa3. (fr onting of the dir ect object ) c. * Aaming ngo I O bei [ ni -bun syu ] DO t I O aa3. (*fr onting of the indir ect object ) d. * Aaming [ ni -bun s yu ] DO ngo I O bei t DO t I O aa3. (*fr onting of both objects ) If the indir ect object cannot undergo object fr onting, ther e is no w a y to cr eate an OS v P : S < DO < I O < V ( cf. ( 303 ) abo v e ), w her e the v erb is put at the right edge in the v P . As such, doubling is the onl y option. 21 20. One possible exp lanation is that the indir ect object is accompanied with a null pr eposition/ dativ e mark er that f orms a larger p hrase with the indir ect object (T ang 1998a ). I t ma y be that the indir ect object is too embedded or object fr onting cannot target a pr epositional p hrase. In either case, the indir ect object is immobile. 21. In a similar v ein, this line of r easoning also rul es out non-doubling cases lik e (i), w her e doubling of ngo is f or ced. I thank an anon ymous JEAL r eview er f or this examp le. (i) Aaming Aaming gaau teach *( ngo ) I jyujinhok linguistics gaa3 sfp ngo . I ‘ Aaming teaches me linguistics. ’ Specificall y , the right dislocation of ngo without doubling r equir es successful fr onting of jyujinhok ‘linguistics’; ho w ev er , it fails to undergo fr onting: (ii) * Aaming Aaming gaau teach jyujinhok i linguistics ngo I t i gaa3. sfp Int.: ‘ Aaming teaches me linguistics. ’ 225 5.4. D ERIVIN G THE AS YMMETRIES IN DO UB LIN G Second, as far as interpr etiv e eff ects ar e concerned, it has been suggested that the object mo v ement that cr eates the ( non-canonical) SO V w or d or der r enders the object a contrastiv e f ocus (Ernst and W a ng 1995 , i.a.). W e observ e a similar interpr etiv e eff ect in case of right dislocation of v erbs w her e ther e is no doubling. In ( 303 a ), ni-di ye ‘this thing’ is contrastiv el y f ocused. The speak er is clarifying the thing that Aaming eats. Cruciall y , such as interpr etiv e eff ect is absent if the v erb is doubled, as in ( 291 a ), w her e the v erb r eceiv es ( contrastiv e ) f ocus interpr etation instead. The speak er is clarifying w hether Aaming r eall y eats the thing or not ( see L. Y .-L. Cheung 2015 ; and also discussions belo w .). W e ar e no w left with the case of right dislocation of a subject without doubling. S imilar to the pr oposed anal y sis f or v erbs abo v e, I suggest that doubling of the subject is not obligatory because the VP can be fr onted to the left edge of v P , r endering the subject on the right edge of the v P . The subject, consequentl y , is right dislocated with its lo w er cop y being deleted b y CD . The sentence in ( 308 ) has the deriv ational history giv en in ( 309 ). (308) N o doubling in right dislocation of subjects Soeng w ant sik eat ni -tiu this-cl jyu fish aa3 sfp A aming . Aaming ‘ Aaming w ants to eat this fish. ’ (309) The schematic deriv ation of RD of subjects VP mo v ement! CD! LIN v P ; OS v P : VP < S [ v P VP S VP ] sfp S Arguments f or this VP mo v ement come in tw o f orms. F irst, VP s can mo v e independentl y of right dislocation. (310 ) [ S ik eat ni -tiu this-cl jyu ] i , fish Aaming Aaming soeng w ant t i aa3. sfp (Lit.) ‘T o eat this fish, Aaming w ants . ’ Second, the VP in ( 308 ) r eceiv es f ocus interpr etation, a discourse eff ect that is extensiv el y dis- 226 5.4. D ERIVIN G THE AS YMMETRIES IN DO UB LIN G cussed in L. Y .-L. Cheung ( 2009 ) ( r ef err ed to as Dislocation F ocus Construction ). If the subject is dou - bled as in ( 295 ), then the it is the subject that r eceiv es the ( contrastiv e ) f ocus interpr etation, instead of the VP . S uch patterns ar e similar to those f ound with the doubling/ non-doubling cases of right dislocation of v erbs, as pr eviousl y discussed. Bef or e I lea v e this subsection, it should be noted that the doubled elements giv e rise to contr astive f ocus instead of inf ormational f ocus. That it is not the latter can be sho wn b y the f ollo wing question- answ er p air: 22 (311) a. A: Aaming Aaming wui will zou do matje w hat aa3? sfp ‘W hat will Aaming do?’ b. B: A aming Aamming wui will tai w atch dinsi TV aa3, sfp A aming . ‘ Aaming will w atch TV . ’ In ( 311 ), A asks about what Aaming will do and thus Aaming cannot be the inf ormational f ocus of an appr opriate answ er . N onetheless, Aaming can be f elicitousl y doubled in the answ er giv en b y B. This suggests that the doubled subject does not bear inf ormational f ocus (i.e. it is the VP ‘w atch TV’ that is inf ormationall y f ocused). In ( 311 b ), it is conceiv able that B is str essing that the answ er onl y app lies to Aaming, but not an y other , pr obabl y because s/he does not kno w about others. 23 T o sum up, the curr ent pr oposal deriv es the doubling asymmetries in right dislocation in a w a y largel y similar to that in topic constructions. The additional comp lication comes fr om the option- ality of doubl ing w hich is onl y observ ed in right dislocation. R ef er encing the occurr ence of v arious independentl y motiv ated v P - internal mo v ements, I suggest that optionality arises as a consequence of w hether these mo v ements occur or not. Cruciall y , again, the exp lanation of doubling asymmetries 22. I thank an anon ymous JEAL r eview er f or raising this point. 23. See L. Y .-L. Cheung ( 2015 ) f or a slightl y diff er ent scenario and mor e discussions on the contrastiv e function of the doubled elements. 227 5.4. D ERIVIN G THE AS YMMETRIES IN DO UB LIN G does not r esort to the p hrase structural status of the mo ving elements. 5.4.3 A r emar k on diff er ences in acceptability As noted in section 5.2 , ther e is a diff er ence in acceptability betw een certain of the unacceptable cases described her e. On one hand, if an element must be doubled in case of mo v ement ( e.g. v erb topical - ization ) but it is not doubled, the sentence is strictl y out ( as in ( 269 )). On the other hand, if an element str ongl y pr ef ers not to be doubled ( e.g. object topicalization and right dislocation ), doubling this el - ements lead to a less sev er e unacceptability ( r elev ant examp les ar e mark ed with ??, such as ( 270 )). In other w or ds, failing to double w hat should be doubled r esults in sentences that ar e mor e sev er el y degraded than instances of doubling w hat should not be doubled, i.e. failing to delete w hat should be deleted. I r ef er to the f ormer cases as f ail- to-double cases and the latter as f ail- to-delete cases . The question is w h y ther e is such a diff er ence. I suggest that the diff er ence indicates that the tw o cases violate diff er ent princip les in the gram- mar . F or the f ail- to-double cases , the absence of the lo w er cop y in v erb topicalization dir ectl y violates the p honological r equir ements imposed b y CL. F or examp le, in deriv ation in ( 287 ), the step in ( 287 b ) dictates that the lo w er cop y of soeng ‘w ant’ must be pr eceded b y Aaming and hai . F ailing to double (i.e. deleting the lo w er cop y ) violates the established OS. V iolation of OS r equir ement leads to a lin- earization failur e and consequentl y the structur e cannot be pr onounced. In contrast, the f ail- to-delete cases do not constitute such a violation. Consider the schematic deriv ation in ( 289 ), failing to delete the lo w est cop y of the object (i.e. the most embedded one ) w ould not lead to failur es in linearization - nothing r estricts the v erb fr om pr eceding the object. Instead, the structur e is degraded due to a failur e to app l y the operation that is r esponsible f or minimizing copies ( e.g. CD ). This ma y violate an econom y princip le in the grammar such as the one giv en belo w: 24 (312) Econom y condition on identical copies Minimize pr onunciation of identical copies. 24. A similar condition is pr oposed in Landau ( 2006 , p.57). 228 5.4. D ERIVIN G THE AS YMMETRIES IN DO UB LIN G So the f ail- to-delete cases violate an econom y condition, rather than r equir ements r elated to lineariza- tion. Assuming that econom y conditions ar e mor e “tolerant” , it ther ef or e leads to a less sev er e lev el of unacceptability . On a r elev ant note, I ha v e pr oposed that the suspension of CD occurs as a last r esort to ensur e successful linearization, in section 5.4.1 . The last r esort natur e can be consider ed as an indication of this econom y condition w hich pr ef ers the app lication of CD , unless its app lication leads to ungrammaticality . 5.4.4 R eso l ving a further as ymmetry in doub ling Bef or e I end this section, I discuss a pr ediction and a further asymmetry in doubling, w hich also f ollo w fr om the curr ent pr oposal. Let us first consider a pr ediction concerning objects. W e ha v e seen that an object cannot be doubled w hen right dislocated (because it is on the right edge ). But if this object also mo v es f or topicalization, then the object mo v ement within v P will establish a diff er ent OS v P : O < S < V . In such case, w e expect to see doubling of the object in right dislocation to be possible. This pr ediction is borne out as seen in ( 313 ). (313) Left -dislocated topics in right dislocation N i -di this-cl je thing Aaming Aaming sik eat aa4 q ni -di this-cl je . thing ‘ Aaming eats THIS THIN G?’ The deriv ation history giv en in ( 314 ) suggests that the object first stops at Spec v P , w hich, after the Spell -Out of v P , mo v es further f or topicalization. The object continues to mo v e f or right dislocation (indicated b y the rightw ar d mo v ement ). In or der to pr eserv e the OS established in v P , the cop y in the topic position must not be deleted b y CD ( such that ther e is a cop y of the object that pr ecedes the subject and the v erb ). The suspension of CD leads to tw o copies in the sentence. (314) The schematic deriv ation of ( 313 ) Obj. mo v e.! CD! LIN v P ; OS v P : O < S < V [ T op P O ... [ v P O S V O ] sfp O ] 229 5.4. D ERIVIN G THE AS YMMETRIES IN DO UB LIN G W hi le ( 313 ) sho w s that a left -dislocated topic can be doubled w hen it further undergoes right dislocation, it should be noted that not all types of topics can be doubled w hen right dislocated. F or examp le, a base generated topic ( e.g. a hanging topic or an aboutness topic ) cannot be doubled w hen it is right dislocated. An examp le of this is giv en in ( 315 ) w her e seoigwo is consider ed as an aboutness topic, as it does not corr espond to a gap in the sentence. (315) Base generated topics in right dislocation (?? Seoigw o ) fruit Aaming Aaming zungji lik e lei pear aa3 sfp seoigw o . fruit ‘ As f or fruits, Aaming lik es pears. ’ The asymmetry betw een a left dislocated topic and a base generated one f ollo w s fr om the curr ent pr oposal. Cruciall y , a base generated topic ha v e a diff er ent deriv ational history to a left -dislocated topic: a base generated topic does not originate within the v P , and thus it is not linearized r elativ e to the elements within v P . Instead, it is base generated in the topic position in the CP domain. W hen it is right -dislocated, CD app lies upon the Spell -Out of the T opic P and it deletes the cop y in the topic position ). As a r esult, a base generated topic is right dislocated without doubling, as schematicall y sho wn in ( 316 ). (316) The schematic deriv ation of ( 315 ) OS v P : S < V < O [ T op P T opic ... [ v P S V O ] sfp T opic ] The curr ent pr oposal thus not onl y captur es the patterns described in section ?? , but it also mak es a pr ecise pr ediction on diff er ent doubling pr ofiles f or diff er ent types of topics w hich f ollo w fr om their deriv ational histories. 230 5.5. AL TERN A TIVE EXP LAN A TI O NS T O THE DO UB LIN G EFFECTS 5.5 Alternati v e e xp lanati ons to the doub ling eff ects In this section, I discuss thr ee alternativ e exp lanations to the general anal y sis of doubling pr oposed her e. F or expository purposes, I f ocus on ho w these alternativ es fail to deriv e the doubling pattern of v erbs in Cantonese, w hich demonstrate the most comp licated pattern f ound (i.e. obligatory doubling and optional doubling). I t should be noted that I am not arguing against the r ole of these alternativ es in potentiall y deriving doubling patterns in other languages, but that they fall short of exp laining the cases w e ha v e seen so far . The first tw o alternativ es str ess the r ole of the p honological component ( similar to the curr ent pr oposal). N unes ( 2004 ) and Corv er and N unes ( 2007 ) tak e adv antage of an independentl y motiv ated operation in the p honological component, namel y , morphological fusion , w hich app lies to tw o adjacent terminals and leads to the f ormation of a comp lex head. This operation is suggested to mak e a mem- ber of a chain to be “in visible” to Cop y Deletion, because it is suggested that a morp hologicall y fused element is no longer identical to its other cop y . T o see ho w such an idea might w ork with v erb topical - ization in Cantonese, it could be the case that the higher cop y of a v erb is morp hologicall y fused with a null T opic head, f orming a comp lex head #V -to p#. Consequentl y , it w ould be r egar ded as distinct fr om the lo w er cop y , i.e. V . CD does not app l y since ther e ar e no identical copies. The same might be suggested f or the doubling case in right dislocation, ex cept that the null head being a diff er ent one, sa y , a head rd, that hosts right dislocated elements in its specifier . H o w ev er , the non-doubling case of v erbs in right dislocation poses a challenge to such an appr oach. T o maintain a morp hological fusion account, one w ould be f or ced to sa y that morp hological fusion is optional. Importantl y , it w ould be onl y optional f or the head rd, but not to p, because w e ha v e seen that doubling in v erb topicalization is obligatory . The optionality of the app lication of morp hological fusion, together with the idiosyncratic natur e of diff er ent functional heads, w eak ens the exp lanatory po w er of such an appr oach. 25 25. F urther challenges ha v e been discussed in Cheng and V icente ( 2013 ), w her e they suggest that it is puzzling that morp hological fusion does not app l y to objects (i.e. no doubling in object topicalization ). C iting an examp le fr om B razilian S ign Language (N unes and Quadr os 2006 ), they note that it is possible to double wh -expr essions (in addition to heads ), 231 5.5. AL TERN A TIVE EXP LAN A TI O NS T O THE DO UB LIN G EFFECTS Another potential alternativ e is pr oposed b y T rinh ( 2009 ). H e pr oposes a constraint on CD , w hich suggests that a lo w er cop y can be deleted onl y if it “ ends” an XP . (317) Constraint on Cop y Deletion ( CCD , T rinh 2009 ) A chain ( , ) is deletabl e onl y if is at the right edge of an XP . App l ying the idea to the Cantonese data, this constraint could captur e the asymmetry betw een v erbs and objects in topicalization: the lo w er cop y of an object is deleted because it ends the v P , w hile that of a v erb is not, because it does not end the v P . Schematicall y , (318) A schematized illustration of v erb and object topicalization in Cantonese V / O ... [ v P S * delete V O K delete O ] This appr oach, ho w ev er , w ould also pr edict that v erb doubling is obligatory w hen the v erb is right dislocated, just lik e the case of v erb topicalization. This is because the v erb is not at the right edge. H o w ev er , as w e ha v e seen, v erb doubling is optional in right dislocation in Cantonese. N ote that object mo v ement within v P (lik e the one w e ha v e seen in § 5.4.2 ) does not help since co v ert elements count in the calculation of w hat “ ends” an XP , accor ding to T rinh ’ s pr oposal (p.195, fn. 18). A thir d possible alternativ e exp lanation might be a r esort to the notion of parallel chains in N arr o w S yntax. F ollo wing Chomsky ( 2008 ), Kandybo wicz ( 2008 ), and Aboh and Dy ak ono v a ( 2009 ), Lai ( 2019 ) pr oposes that doubling is due to the cr eation of parallel chains. The idea is that an element mo v es to tw o higher positions, cr eating tw o independent chains that ha v e the same tail (i.e. the lo w er cop y ). W hen CD app lies, onl y the lo w er cop y is deleted, since both higher copies surviv e CD and hence ther e is doubling. W hile Lai primaril y discusses the doubling cases in right dislocation, the same r easoning w hich pr esumabl y ha v e a comp lex internal structur e (Cable 2007 ). W or ds in capital letters indicate the glosses f or B razilian S ign Language. (i) J O HN SEE WH O YESTERD A Y WH O ‘W ho exactl y is it that J ohn sa w y ester da y ?’ 232 5.6. EXTENSI O N: VERB M O VEMENT WITH O UT DO UB LIN G might be used to app l y to v erb doubling as f ollo w s. In v erb topicalization, a v erb might be suggested to mo v e independentl y to both the v head and the T opic head, cr eating tw o separate chains. W hen CD app lies, onl y the V head is deleted, r esulting in tw o occurr ences of the v erb in the v and T opic heads. (319) A schematized illustration of parallel chains in v erb topicalization in Cantonese [ T opic P V - T opic ... [ v P S V - v [ VP V O ]] Analogousl y , parallel chains might also be able to account f or v erb doubling in right dislocation, if w e assume the v erb mo v es to a counterpart of the T opic head that hosts right dislocation. H o w ev er , this appr oach does not pr edict the optionality of doubling in right dislocation. T o account f or the absence of doubling in right dislocation, one must suggest that the higher cop y of the chain f V - v , Vg can sometimes be deleted. S uch deletion is technicall y challenging under a parallel chain appr oach as w e ha v e to stipulate an uncon v entional deletion operation that targets the w hole chain. In sum, I conclude that existing accounts of doubling primaril y f ocus on ho w to ensur e the second occurr ence of a cop y , but they ar e less adaptable to the optional natur e of v erb doubling in Cantonese. F urther challenges to these alternativ e accounts ar e posed b y the doubling pattern of objects and sub- jects w her e doubling is not strictl y pr ohibited or r equir ed. The curr ent pr oposal, b y w a y of contrast, off ers a mor e compr ehensiv e account on the doubling patterns in Cantonese. 5.6 Extensi on: v erb mo v ement without doub ling Thus far , w e ha v e f ocused ex clusiv el y on data in Cantonese. W e no w turn to some additional cr oss- linguistic pattern. Giv en the curr ent exp lanation of ( v erb ) doubling, the r elativ e position of the v erb to the subject is al w a y s fix ed at the Spell -Out of v P . The pr oposal seems to pr edict that the w or d or der of a langu age is e ither S- V -O (i.e. the v erb does not mo v e acr oss the subject in the v P ) or V -S- V -O (i.e. the v erb mo v es acr oss the subject and is doubled). This is too str ong in the sense that it (incorr ectl y ) rules out any V -S or der . F or examp les, sentence s in ( 320 ) sho w tw o cases of v erb mo v ement acr oss 233 5.6. EXTENSI O N: VERB M O VEMENT WITH O UT DO UB LIN G the subject without doubling in S w edish and Bulgarian. These non-doubling cases do not immediatel y f ollo w fr om the pr oposal put f orth in § 5.3 . (320 ) V erb mo v ement without doubling a. S w edish Hittade f ound han he faktist actuall y (*hittade ) f ound pengarna money .the under under sängen? bed.the ‘Did he actu all y find the money under the bed?’ (T akita 2010 , p.40, with adaptations ) b. Bulgarian Razkazv ala told beše w as često often M arija M aria (*razkazv ala ) told tazi this istorija story ‘M aria had often told this story . ’ (H arizano v 2019 , p.8, with adaptations ) These cases, together with the Cantonese data, sho w that language s v ary with r egar d to w hether mo v ement of v erbs o v er the subject r equir e doubling of the v erbs or not. F ollo wing T akita ( 2010 ), I suggest this cr oss- linguistic diff er ence r esults fr om the parameter of Spell -Out domain. Specifi - call y , w hile Spell -Out in v ariabl y app lies at v P , languages ma y diff er in the size of the linearization domain. 26 I suggest that languages lik e S w edish and Bulgarian ha v e a diff er ent linearization domain fr om Cantonese. F or these languages, upon Spell -Out, onl y the comp lement of v but not the w hole v P is lin earized. (321) Spell -Out Domain P arameter f or v P (T akita 2010 ) W hen Spell -Out app lies to v P , a. Linearize the w hole v P , including the elements on its edge, or b. Linearize the comp lement of v . If S w edish and Bulgarian tak e the v alue of ( 321 b ), w hen Spell -Out app lies to the v P , onl y the VP is linearization ( as opposed to the v P in Cantonese ). Consequentl y , the or der betw een S and V is not 26. The idea that the linearization domain of v P v aries acr oss languages has its r oots in K o ( 2005 , 2007 ), w ho pr oposes that the linearization domain of K or ean is v P , rather than VP ( w hich is assumed to be the linearization domain f or some Scandina vian languages; see also discussions in F o x and P esetsky ( 2005 , §5). 234 5.6. EXTENSI O N: VERB M O VEMENT WITH O UT DO UB LIN G fix ed upon Spell -Out of v P . This is illustrated with the S w edish examp le in ( 320a ). ( 322 b ) onl y giv es the OS upon the the first instance of Spell -Out, w her e the OS onl y contains or dering inf ormation of V , O and P P , to the ex clusion of S. The v erb is thus fr ee to mo v e at a later stage of the deriv ation. The same app lies to Bulgarian data as w ell. (322) The deriv ation of ( 320a ) a. [ CP Hittade [ TP han faktist [ v P han [ VP hittade pengarna under sängen? ]]] b. LIN VP ; OS VP : V < O < P P If this line of r easoning is on the right track, the a v ailability of v erb doubling is corr elated with the size of Spell -Out domain in the f ollo wing w a y: (323) V erb doubling possibility a. e.g. Cantonese Languages that allo w v erb doubling tak e the v alue of ( 321 a ); b. Languages that disallo w v erb do ubling tak e the v alue of ( 321 b ). e.g. S w edish, Bulgarian Inter estingl y , the parameter in ( 321 ) is originall y pr oposed in T akita ( 2010 ) to exp lain illicit cases of r emnant mo v ements in J apanese and licit ones in English and German. Specificall y , he pr oposes that languages that tak e the v alue of ( 321 a ), i.e. v P is linearized upon Spell -Out, w ould disallo w r emnant mo v ement. This describes the case of J apanese. Consider the f ollo wing paradigm ( adapted fr om T akita ( 2010 , p.11-12)). ( 324 a ) is the baseline. ( 324 b ) sho w s that long distance scrambling of P P s is possible. ( 324 c ) sho w s that CP scrambling is also possible. ( 324 d), ho w ev er , sho w s that once a P P is scrambled out of an embedded CP , the ( r emnant ) CP cannot be scrambled. (324) a. Baseline T a r oo-ga T a r oo-n o m [ CP H anak o-ga H anak o-n o m [ P P Sooru - ni Seoul - in ] i - ru be-pres to that ] omottei - ru. think -pres ‘T ar oo thinks [ that H anak o liv es [ in Seoul] ]. ’ 235 5.6. EXTENSI O N: VERB M O VEMENT WITH O UT DO UB LIN G b. (Long distance ) P P scrambling [ P P Sooru - ni Seoul - in ] i , T ar oo-ga T ar oo-n o m [ CP H anak o-ga H anak o-n o m t i i - ru be-pres to that ] omottei - ru. think -pres ‘(lit.) [In Seoul] i , T ar oo thinks [ that H anak o liv es t i ]. ’ c. CP scrambling [ CP H anak o-ga H anak o-n o m [ P P Sooru - ni Seoul - in ] i - ru be-pres to that ] j T ar oo-ga T ar oo-n o m t j omottei - ru. think -pres ‘(lit.) [That H anak o liv es [ in Seoul] ] j , T ar oo thinks t j . ’ d. P P scrambling f ollo w ed b y CP scrambling * [ CP H anak o-ga H anak o-n o m t i i - ru be-pres to that ] j [ P P Sooru - ni Seoul - in ] i T ar oo-ga T ar oo-n o m t j omottei - r u. think -pres ‘(lit.) [That H anak o liv es t i ] j , [ in Seoul] i , T ar oo thinks t j . ’ The unacceptability of ( 324 d) has been attributed to v ersions of the Pr oper B inding Condition (P BC, F iengo 1977 ; Saito 2003 ), w hich states that traces must be bound. T akita ( 2010 ) casts doubt on the pr ecise natur e of the P BC in the grammar and instead pr oposes that the P BC-eff ects observ ed abo v e f ollo w fr om some general princip le concerning linearization. A dopting the idea of C y clic Lineariza- tion, he pr oposes that r emnant scrambling is ruled out because it leads to conflicts in linearization. T o see ho w , consider first the (long distance ) scrambling of the P P in ( 324 d), w hich r equir es the P P to mo v e to the edge of the v P . S ince J apanese tak es the v alue of ( 321 a ), the w hole v P is linearized, giving the OS v P : PP < S < V . The P P is subsequentl y further scrambled out of the CP . Then, the r emnant CP is scrambled to a position higher than the P P . This establishes another OS at the final Spell -Out: CP < PP . N ote that this CP contains the subject and the v erb w hose or der is r elativized to the P P alr eady . So the scrambling of CP w ould giv e rise to OS as f ollo w s: S < V CP < PP . This r esults in linearization conflicts and hence causes the unacceptability of ( 324 d). R emnant mo v ement/ scrambling in J apanese is thus sy stematicall y ruled out b y C y clic Linearization and the supp lementary assumption that the linearization domain of J apanese is v P . 236 5.6. EXTENSI O N: VERB M O VEMENT WITH O UT DO UB LIN G N o w consider r emnant mo v ement in English and German. In contrast to J apanese, these lan- guages allo w s r emnant mo v ement. (325) Licit r emnant cases a. English: A -movement f ollowed by r emnant v P movement [ C riticized t i b y his boss ] j , J ohn i has nev er been t j . b. German: object scr ambling f ollowed by r emnant v P /VP topicalization [ t i Z u to lesen ] j r ead hat has k einer no.one [ das the Buch ] i book t j v ersucht. tried ‘N o on e has tried to r ead the book’ T a kita ( 2010 ) suggests that English and German ha v e a diff er ent linearization domain, w her e onl y the elements in the VP ar e linearized upon Spell -Out of v P . T ak e the English case in ( 325 a ) as an examp le. Assuming that the object J ohn can mo v e to the edge of v P due to passivization, its linear or der with r egar d to other elements is not fix ed upon the Spell -Out of v P . This is because onl y the elements in VP ar e linearized in English, w hich giv es rise to the OS v P : V < PP . F ollo wing this, the object J ohn undergoes further mo v ement to Spec TP . S ubsequent ( r emnant ) mo v ement of the v P w ould not r esult in linearization conflicts: no OS f orbids V or P P fr om pr eceding the object. As a r esult, r emnant mo v ement is allo w ed in languages with a linearization domain of VP under C y clic Linearization. The cases of J apanese and English/ German can be summarized as f ollo w s: the a v ailability of r em- nant mo v ement is corr elated with the size of the Spell -Out domain in a w a y specified in ( 326 ). (326) R emnant mo v ement possibility a. Languages that disallo w r emnant mo v ement tak e the v alue of ( 321 a ); e.g. J apanese b. Languages that allo w r emnant mo v ement tak e the v alue of ( 321 b ). e.g. German, English R eturning to our discussion on v erb doubling, ( 326 ) and ( 323 ) combine to pr edict the distribution 237 5.6. EXTENSI O N: VERB M O VEMENT WITH O UT DO UB LIN G of v erb doubling and r emnant mo v ement, since they r equir e a language to tak e diff er ent v alues on the Spell -Out domain parameter suggested in ( 321 ). The pr edictions ar e giv en in ( 327 ): (327) Pr edicted distribution of v erb doubling and r emnant mo v ement a. Languages that allo w v erb doubling will disallo w r emnant mo v ement. b. Languages that allo w r emnant mo v ement will disallo w v erb doubling. ( 327 a ) is borne out b y the una v ailability of r emnant v P mo v ement in Cantonese. ( 328a ) sho w s that v P - fr onting is disallo w ed if the subject is mo v ed out fr om the v P , i.e. v P fr onting is disallo w ed in raising constructions. This f ollo w s the same line of r easoning as the exp lanation of w h y r emnant scrambling is disallo w ed in J apanese. N ote that ( 328b ) is supp lied to sho w that v P - fr onting is allo w ed if the subject of the v P is not mo v ed out, i.e. v P fr onting is allo w ed in contr ol constructions. (328) v P fr onting in Cantonese a. Raising * [ v P t i bin become hak dark ] j , go cl tin i sky hoici begin t j laa3. sfp Intended: ‘T o become dark, the sky begins. ’ b. Contr ol [ v P P R O pau run coengpau long.run ] i , k eoi k eoi soengsi -gw o try -exp t i laa3. sfp ‘T o run long distance, he tried. ’ ( 327 b ) is borne out in English. W hile r emnant mo v ement is allo w ed, v erb doubling is not, as in ( 329 ). W ith a smaller Spell -Out domain, V occupies the edge position and is thus fr ee to mo v e. Doubling is not r equir ed, hence disallo w ed ( due to the last r esort natur e ). 27 27. I t should be noted that mo v ement of the v erb without doubling is also disallo w ed : (i) * Criticized, J ohn t i his boss. This should not be r egar ded as a counter examp le to the curr ent pr oposal, since, the curr ent pr oposal states the necessary condition f or doubling, but not the sufficient condition f or doubling. W hile v erb mo v ement without doubling is allo w ed in S w edish and Bulgarian, sentences lik e (i) in English ma y be ruled out on independent gr ounds. I do not pursue this further in the curr ent chapter . 238 5.6. EXTENSI O N: VERB M O VEMENT WITH O UT DO UB LIN G (329) V erb doubling in English * Criti cize( d) , J ohn criti cized his boss. As a final r emark, I briefl y discuss w hat the curr ent pr oposal does not necessaril y pr edict, listed in ( 330 ). The curr ent pr oposal concerns the necessary condition f or doubling. If a language disallo w s v erb doubling, it need not allo w r emnant mo v ement, as ther e ma y be independent r easons to rule out v erb doubling. Lik ewise, if a language disallo w s r emnant mo v ement, it need not allo w v erb doubling either . (330 ) Some non-pr edictions a. Languages that disallo w s v erb doubling will allo w r emnant mo v ement. b. Languages that disallo w s r emnant mo v ement will allo w v erb doubling. I t is, ho w ev er , inter esting to see that if a language disallo w s v erb doubling pr ecisely because of its Spell - Out domain being a VP , w e do expect to see r emnant mo v ement to be possible. This has alr eady been seen in English and w e also observ e r emnant VP topicalization in S w edish, as in ( 331 ). (331) R emnant mo v ement in S w edish (F o x and P esetsky 2005 , p.25) ? [ Gett giv en henne her t i ] har ha v e jag I den i it inte not ... ‘I ha v e not giv en it to her . ’ W hether ( 330 b ) holds is less clear , ho w ev er . T o the best of m y kno w ledge, the closest J apanese and K or ean counterparts of the Cantonese v erb doubling constructions ar e discussed in N ishiy ama and Cho ( 1998 ), w her e both languages disp la y some doubling eff ects in pr edicate cleft constructions. (332) J apanese J ohn-ga J ohn-n o m computer -o computer -a cc kai - w a buy -co n si -ta do-pst ‘Indeed, J ohn bought a computer , (but...)’ 239 5.7. CO N CL USI O NS (333) K or ean J ohn- i J ohn-n o m computer - lul computer -a cc sa - ki - nun buy -ki-co n sa -ss-ta buy -p ast-d ecl ‘Indeed, J ohn bought a computer , (but...)’ In the J apanese case, the v erb is associated with a dumm y v erb instead of an identical cop y , w her eas the K o r ean case comes closer to a case of v erb doubling, but the first v erb is mark ed with the morp heme -ki . F or r easons of space, I will lea v e the full in v estigation of these examp les to futur e r esear ch. T able 5.2 summarizes the findings in this section. P arameter ( 321 a ) Linearize v P ( 321 b ) Linearize VP Language Cantonese J apanese K or ean S w edish English German V erb doubling Y es ? ? N o N o N o R emnant mo v ement N o N o N o Y es Y es Y es T able 5.2: V erb doubling and r emnant mo v ement acr oss languages S u mming up, this section began with cases of v erb mo v ement without doubling in languages lik e S w edish and Bulgarian. F ollo wing T akita ( 2010 ), I pr oposed that the diff er ence betw een languages with/ without v erb doubling lies in the Spell -Out domain parameter giv en in ( 321 ). I then suggested that such an exp lanation is further corr oborated b y a corr elation betw een v erb doubling and r emnant mo v ement, the a v ailability of w hich depends on the parametric v alue f or ( 321 ) that the language tak es. 5.7 Conclusi ons In this chapter , I began with a consideration of an asymmetry in v erb topicalization and object topi - calization. I set up the empirical f oundation of this study b y giving a description of v arious patterns of doubling in Cantonese. I discussed the doubling pr ofiles of v erbs, subjects and objects with r egar d to topic constructions and right dislocation. I then pr oposed an account based on C y clic Linearization. Specificall y , I pr oposed that doubling is a consequence of the suspension of Cop y Deletion. Cop y 240 5.7. CO N CL USI O NS Deletion is suspended as a last r esort to a v oid violations of linearization r equir ements im posed b y CL. In the final section, I discussed cases w her e v erb mo v ement does not disp la y doubling eff ects in languages other than Cantonese, and suggested an account based on the parameter of the linearization domain, f ollo wing ideas in T akita ( 2010 ). The imp lications of the curr ent pr oposal ar e tw o- f old. F irst, it lends further support to C y clic Linearization, w hich has been argued to captur e diff er ent p henomena in diff er ent languages, e.g. ob- ject shift in Scandina vian languages (F o x and P esetsky 2005 ), quantifier floating in K or ean (K o 2005 , 2007 ), r emnant mo v ement in J apanese (T akita 2010 ), pr eposition stranding in English (Drummond, H ornstein, and Lasnik 2010 ), constraints on the scrambling of genitiv e- mark ed arguments in K o- r ean (S impson and P ark 2019 ) and intermediate stranding in a number of languages (Da vis 2020 ). CL str esses the r ole of the p honological component in the study of syntactic locality , serving as an alter - nativ e dir ection to Chomsky’ s v ersion of p hase theory (Chomsky 2000 , 2001 ), one that suggests that a s yntactic domain is inaccessible both to syntactic and p honological operations. Second, the pr oposal deriv es the Cantonese doubling pattern without linking t his to the p hrase- structural status of the ( non-)doubling elements and maintains that the mechanism that determines cop y pr onunciation is the same f or heads and p hrases. This r esonates with r ecent eff orts in unifying head and p hrasal mo v ement. F or examp le, i t is argued that all mo v ement operations lea v e a trace that f eed interpr etation (H artman 2011 ); substitution, in addition to adjunction, is a v ailable to both head and p hrasal mo v ement (F unak oshi 2012 , 2014 ); head mo v ement can target specifier positions just lik e p hrasal mo v ement (H arizano v 2019 ; H arizano v and Gribano v a 2019 ); and dependencies betw een arguments, non-arguments, and heads ma y lead to structur e r eduction in the f ormation of infinitiv al clauses (P esetsky 2020 ). This findings of this chapter pr o vide a further piece of evidence along such lines. 241 Chapter 6 Conclusi ons Based on in-depth in v estigations into v arious cases of v erb disp lacement in Cantonese, it is hoped that this thesis contributes to our understanding of mo v ement theories of natural language. F ollo wing up on the debates of the theor etical status and empirical pr operties of head mo v ement, I exp lor ed the possibility of a unified theory of mo v ement that does not mak e r ef er ence to structural types such as heads and p hrases. I discussed thr ee pieces of evidence fr om Cantonese, sho wing that mo v ement of heads and p hrases ar e subject to the same set of syntactic princip les, w hich constrain (i) ho w they mo v e in the syntax, (ii) ho w they contribute to interpr etation, and (iii) ho w their chains ar e p honologicall y r ealized. T o the extent that head mo v ement can be assimilated to p hrasal mo v ement, this thesis sets the basis of a mo v ement theory that does not discriminate heads fr om p hrases, hence a unified theory of mo v ement. T o r estate the theor etical consequences of a unified theory of mo v ement, first, it allo w s us to main- tain the f ormulation of the structur e-building operation, M erge , in its simp lest f orm. Internal M erge app lies to syntactic constituents without the need to distinguish heads fr om p hrases, in a w a y com- parable to External M erge, w hich app lies equall y to both heads and p hrases. Second, it opens up questions of w hether and ho w other r eported diff er ences betw een mo v ement of heads and p hrases can b e attributed to components of the grammar other than the mo v ement mechanism. 242 REFEREN CES R ef er e nces Abels, Klaus. 2001. “The pr edicate cleft construction in R ussian. ” In Pr oceedings o f F ormal Appr oaches to Slavic Linguistics 9 , edited b y S tev en F ranks, T racy H ollo w a y King, and Michael Y adr off, 1–18. Ann Arbor: Michigan Sla vic Publications. Abels, Klaus. 2003. “S uccessiv e C y clicity , Anti - locality , and A dposition S tranding. ” PhD diss., U niv er - sity of Connecticut. Aboh, Enoch O ladé, and M arina Dy ak ono v a. 2009. “Pr edicate doubling and parallel chains. ” Lingua 119 (7): 1035–1065. A dger , Da vid. 2013. A syntax o f substance. Cambridge: MIT Pr ess. A elbr echt, Lobk e, and M ar cel den Dikk en. 2013. “Pr eposition doubling in Flemish and its imp lications f or the syntax of Dutch P P s. ” J ournal o f Compar ative Germanic Linguistics 16 (1): 33–68. Aissen, J udith. 1974. “V erb raising. ” Linguistic I nquiry 5 (3): 325–366. Antonenk o, Andr ei. 2019. “Pr edicate Doubling in R ussian: One pr ocess or tw o?” In Pr oceedings o f F ASL 27 . Michigan Sla vic Pu blications. A oun, J osep h, and Y en- H ui A udr ey Li. 2003. Essays on the r epr esetntational and derivational natur e o f gr ammar . Cambridge, MA: The MIT Pr ess. 243 REFEREN CES A oun, J osep h, and Y en- H ui A udr ey Li. 2008. “Ellipsis and missing objects. ” In F oundational I ssues in Linguistic Theory, edited b y R obert F r eidin, Carlos P . Oter o, and M aria Luisa Z ubizarr eta, 251– 274. Cambridge, MA: MIT Pr ess. A o y agi, Hir oshi, and T oru I shii. 1994. “On NP I Licensing in J apanese. ” In J apanese/K or ean Linguis - tics(4), edited b y N orik o Akatsuka, 295–312. S tanf or d: S tanf or d Linguisti cs Association. Arr egi, Karlos, and Asia Pietraszk o. 2021. “The U ps and Do wns of H ead Disp lacement. ” Linguistic I n- quiry 52 (2): 241–290. Badan, Linda. 2007. “High and lo w perip hery: a comparison betw een I talian and Chinese. ” PhD diss., U niv ersita’ Degli S tudi di P ado v a. Bak er , M ark. 1985. “The Mirr or Princip le and M orp hosyntactic Exp lanation. ” Linguistic I nquiry 16 (3): 373–416. Bak er , M ark. 1988. I ncorpor ation. Chicago, Illinois: U niv ersity of Chicago Pr ess. Barrie, Michael, and Eric M athieu. 2016. “N oun incorporation and p hrasal mo v ement. ” N atur al Lan- guage and Linguistic Theory 34 (1): 1–51. Barwise, J ohn, and R obin Cooper . 1981. “Generalized Quantifiers and N atural Language. ” Linguistics and Philosophy 4:159–219. Bastos-Gee, Ana C. 2009. “T opicalization of v erbal pr ojections in B razilian P ortuguese. ” In Minimalist essays on Br azilian P ortuguese S yntax, edited b y J air o N unes, 161–190. Amster dam: J ohn Ben- jamins. Beck, S igrid. 1996. “W h-constructions and transpar ent Logical F orm. ” PhD diss., U niv ersität T übin- gen. 244 REFEREN CES Beck, S igrid. 2006. “Interv ention Eff ects F ollo w F r om F ocus Interpr etation. ” N atur al Language Se - mantics 14 (1): 1–56. Beck, S igrid, and Shin-Sook Kim. 1997. “On WH - and operator scope in K or ean. ” J ournal o f East Asian Linguistics 6:339–384. Beck, S igrid, and Arnim v on S techo w . 2015. “E v ents, T imes and W orlds - An LF Ar chitectur e. ” In Situationsargumente im N ominalber eich, edited b y F ortmann v on Christian, 13–46. Berlin: de Gruyter . Benedicto, Elena. 1998. “V erb M o v ement and its Eff ects on Determinerless Plural S ubjects. ” In R o- mance Linguistics : Theor etical P erspectives, edited b y Armin Sch w egler , Bernar d T ranel, and M yr - iam U ribe- E tx ebarria, 25–40. Amster dam: J ohn Benjamins. Benincà, P aola, and Cecilia P oletto. 2004. “ A Case of Do-S upport in R omance. ” N atur al Language and Linguistic Theory 22 (1): 51–94. B hatt, Rajesh. 1998. “Obligation and P ossession. ” In P apers fr om the UP enn/MIT R oundtable on Argu- ment S tructur e and Aspect, edited b y H eidi H arley , 21–40. Cambridge, MA: MITWP L. B lanchette, F rances, and Chris Collins. 2019. “On the subject of negativ e auxiliary in v ersion. ” Cana- dian J ournal o f Linguistics 64 (1): 32–61. Bobaljik, J onathan. 2002. “ A -chains at the P F - interface: copies and ‘ co v ert’ mo v ement. ” N atur al Lan- guage and Linguistic Theory 20 (2): 197–267. Bobaljik, J onathan Da vid, and Samuel B r o wn. 1997. “Interarbor eal Operations: H ead M o v ement and the Extension R equir ement. ” Linguistic I nquiry 28 (2): 345–356. 245 REFEREN CES Bobaljik, J onathan Da vid, and S usi W urmbrand. 2012. “W or d Or der and Scope: T ranspar ent Interfaces and the ¾ S ignatur e. ” Linguistic I nquiry 43 (3): 371–421. Boeckx, Cedric, and Sandra S tjepano vić. 2001. “H ead- ing to w ar d P F . ” Linguistic I nquiry 32:345–355. Borsley , R obert D , M aria- Luisa Riv er o, and J anig S tep hens. 1996. “Long head mo v ement in B r eton. ” In The S yntax o f the Celtic Languages: A Compar ative P erspective, edited b y R obert D Borsley and I an R oberts, 53–74. Cambridge: Cambridge U niv ersity Pr ess. Bošk o vić, Željk o. 2007. “On the Locality and M otiv ation of M o v e and Agr ee: An E v en M or e Minimal Theory . ” Linguistic I nquiry 38 (4): 589–644. Bošk o vić, Željk o. 2014. “N o w I’ m a Phase, N o w I’ m N ot a Phase: On the V ariability of Phases with Extraction and Ellipsis. ” Linguistic I nquiry 45 (1): 27–8 9. Bošk o vić, Željk o, and J air o N unes. 2007. “The Cop y Theory of M o v ement: A view fr om P F . ” In The Copy Theory o f M ovement (Linguistics T oday 107), edited b y J air o N unes and N orbert H ornstein, 13–74. O xf or d: B lackw ell Publishers. B ranan, K en y on, and Michael Y oshitaka Erlewine. 2020. Anti-pied-piping. M s., N ational U niv ersity of S ingapor e. B rattico, P auli. 2021. “Pr edicate clefting and long head mo v ement in F innish. ” Linguistic I nquiry: 1–57. B r ody , Michael. 2000. “Mirr or Theory: syntactic r epr esentation in perf ect syntax. ” Linguistic I nquiry 31 (1): 29–56. Büring, Daniel. 1997. The M eaning o f T opic and F ocus: The 59th S tr eet Bridge A ccent R outledge S tudies in German Linguistics. London and N ew Y ork: R outledge. Bury , Dirk. 2003. “Phrase S tructur e and Deriv ed H eads. ” PhD diss., U niv ersity College London. 246 REFEREN CES Cable, Seth. 2004. Pr edicate Clefts and Base -Gener ation: E vidence Fr om Yiddish and Br azilian P ortuguese. M s., MIT , Cambridge, MA. Cable, Seth. 2007. “The Grammar of Q: Q- P articles and the N atur e of W h- F r onting, as R ev ealed b y the W h-Questions of Tlingit. ” PhD diss., M assachusetts Institute of T echnology . Cable, Seth. 2010. The Gr ammar o f Q: Q- P articles, W h- M ovement and Pied- Piping. O xf or d: O xf or d U niv ersity Pr ess. Chan, B rian H ok -Shing. 2013. “Sentence- final particles, comp lementizers, antisymmetry , and the F inal - o v er - F inal Constraint. ” In Theor etical Appr oaches to Disharmonic W or d Or der , edited b y Ther esa B iberauer and Michelle Sheehan, 445–468. O xf or d: O xf or d U niv ersity Pr ess. Chan, K wun Kin. 2016. A study o f sentence - final phr asal r eduplication in Cantonese. MA thesis, The Chinese U niv ersity of H ong K ong. Chao, Y R. 1968. A Gr ammar o f Spoken Chinese. Berk eley: U niv ersity of Calif ornia Pr ess. Cheng, Lisa Lai Shen. 2008. “Deconstructing the shì ... De construction. ” Linguistic R eview 25:235–266. Cheng, Lisa Lai -Shen. 1991. “On the T ypology of WH -Questions. ” PhD diss., M assachusetts Institute of T echnology . Cheng, Lisa Lai -Shen, and Luis V icente. 2013. “V erb doubling in M andarin Chinese. ” J ournal o f East Asian Linguistics 22 (1): 1–37. Cheng, S iu P ong. 2015. “The R elationship of S yntactic and Semantic Aspects of P ostv erbal P articles and Their Pr ev erbal Counterparts in H ong K ong Cantonese. ” PhD diss., The Chinese U niv ersity of H ong K ong. 247 REFEREN CES Cheung, Candice Chi - H ang. 2008. “W h- fr onting in Chinese. ” PhD diss., U niv ersity of Southern Cali - f ornia. Cheung, Candice Chi - H ang. 2015. “On the F ine S tructur e of the Left P erip hery . ” In The Cartogr aphy o f Chinese S yntax, edited b y W ei - T ien Dy lan T sai, 75–130. O xf or d: O xf or d U niv ersity Pr ess. Cheung, La wr ence Y am- Leung. 1997. A study o f right dislocation in Cantonese. MA thesis, The Chinese U niv ersity of H ong K ong. Cheung, La wr ence Y am- Leung. 2005. “S yntax and semantics of dislocation f ocus construction in Can- tonese. ” PhD diss., U niv ersity of Calif ornia, Los Angeles. Cheung, La wr ence Y am- Leung. 2009. “Dislocation f ocus construction in Chinese. ” J ournal o f East Asian Linguistics 18 (3): 197–232. Cheung, La wr ence Y am- Leung. 2015. “B i -clausal sluicing appr oach to dislocation cop ying in Can- tonese. ” I nternational J ournal o f Chinese Linguistics 2 (2): 227–272. Chomsky , N oam. 1957. S yntactic structur es. M outon. Chomsky , N oam. 1973. “Conditions on T ransf ormations. ” In A F estschrift f or M orris H alle, edited b y S tep hen Anderson and P aul Kiparsky , 232–286. N ew Y ork: H olt Rinehart / W inston. Chomsky , N oam. 1977. “Conditions on transf ormations. ” In Essays on f orm and interpr etation, 81–162. N ew Y ork, N ew Y ork: Elsevier N orth- H olland, Inc. Chomsky , N oam. 1981. Lectur es on government and binding. Dor dr echt, The N etherlands: F oris Pub- lications. Chomsky , N oam. 1986. Barriers. Cambridge, M assachusetts: MIT Pr ess. 248 REFEREN CES Chomsky , N oam. 1994. Bar e phr ase structur e. Cambridge, MA: MITWP L. Chomsky , N oam. 1995a. “Bar e p hrase structur e. ” In Government binding theory and the minimalist pr o- gr am, edited b y Gert W ebelhuth, 383–439. O xf or d: O xf or d U niv ersity Pr ess. Chomsky , N oam. 1995b. The minimalist pr ogr am. Cambridge, M assachusetts: MIT Pr ess. Chomsky , N oam. 2000. “Minimalist inquiries: the framew ork. ” In S tep by step: Essays on minimalist syntax in honor o f H owar d Lasnik, edited b y R oger M artin, Da vid Michaels, and J uan U riager eka, 89–156. Cambridge, MA: MIT Pr ess. Chomsky , N oam. 2001. “Deriv ation b y p hase. ” In K en H ale: a lif e in language, edited b y Michael K en- sto wicz, 1–52. Cambridge, MA: MIT Pr ess. Chomsky , N oam. 2008. “On Phases. ” In F oundational I ssues in Linguistic Theory: Essays in H onor o f J ean- R oger V ergnaud, edited b y R obert F r eidin, Carlos Oter o, and M aria Luisa Z ubizarr eta, 133– 166. Cambridge: MIT Pr ess. Chou, Chao-ting T im. 2013. “U n v alued interpr etable f eatur es and topic A - mo v ement in Chinese rais- ing modal constructions. ” Lingua 123:118–147. C inque, Guglielmo. 1999. Adverbs and functional heads: a cr oss -linguistic perspective. N ew Y ork: O xf or d U niv ersity Pr ess. Collins, Chris. 1997. Local economy. Cambridge, MA: MIT Pr ess. Constant, N oah, and Chloe Gu. 2010. “M andarin ‘ ev en ’ , ‘ all’ and the T rigger of F ocus M o v ement. ” U niversity o f P ennsylvania W orking P apers in Linguistics 16 (1): 4. Corv er , N orbert, and J air o N unes. 2007. The Copy Theory o f M ovement. Amster dam/Philadelp hia: J ohn Benjamins Publishing Compan y . 249 REFEREN CES Da vis, Colin. 2020. “Cr ossing and stranding at edges: On intermediate stranding and p hase theory . ” Glossa: a journal o f gener al linguistics 5 (1): 1–32. Dékán y , É v a. 2018. “ Appr oaches to head mo v ement: A critical assessment. ” Glossa: a journal o f gener al linguistics 3 (1): 1–43. den Besten, H ans. 1983. “On the Interaction of R oot T ransf ormations and Lexical Deletiv e R ules. ” In On the F ormal S yntax o f the W estgermania, edited b y W Abraham. Amster dam, The N etherlands: J ohn Benjamins Publishing Compan y . den Besten, H ans, and Gert W ebelhuth. 1990. “S tranding. ” In Scr ambling and Barriers, edited b y Gün- ther Gr ew endorf and W olf gang S ternef eld, 77–92. Amster dam: J ohn Benjamins Publishing Com- pan y . den Dikk en, M ar cel. 2006. R elators and Linkers. Cambridge, MA: MIT Pr ess. Donati, Caterina. 2006. “On W h - head mo v ement. ” In W h-movement: M oving On, edited b y Lisa Lai - Shen Cheng and N orbert Corv er , 21–46. Cambridge, MA: MIT Pr ess. Drummond, Alex, N orbert H ornstein, and H o w ar d Lasnik. 2010. “ A Puzzle about P -S tranding and a P ossible Solution. ” Linguistic I nquiry 41 ( 4): 689–692. Embick, Da vid, and R oum y ana Izv orski. 1997. “P articip le- A uxiliary W or d Or ders in Sla vic. ” In F ormal Appr oaches to Slavic Linguistics (F ASL ): The Cornell M eeting 1995, edited b y N atasha K ondrasho v a, W a y les B r o wne, E w a Dornisch, and Draga Zec, 4:210–239. Ann Arbor: Michigan Sla vic Publi - cations. Embick, Da vid, and R olf N o y er . 2001. “M o v ement operations after syntax. ” Linguistic I nquiry 32 (4): 555–595. 250 REFEREN CES Emonds, J osep h E. 1970. “R oot and S tructur e- Pr eserving T ransf ormations. ” PhD diss., M assachusetts Institute of T echnology . Emonds, J osep h E. 1976. A T r ansf ormational Appr oach to English S yntax. N ew Y ork, N ew Y ork: A ca- demic Pr ess. Emonds, J osep h E. 1978. “The V erbal Comp lex V’ - V in F r ench. ” Linguistic I nquiry 9:151–175. Erlewine, Michael Y oshitaka. 2015. I n def ense o f Closeness: f ocus -sensitive adverb placeme nt. Erlewine, Michael Y oshitaka. 2017. “W h y the null comp lementizer is special in comp lementizer - trace eff ects. ” In A pesky set: P apers f or David P esetsky, edited b y Clair e H alpert, H adas K otek, and Coppe v an U rk, 371–380. MIT W orking P apers in Linguistics. Erlewine, Michael Y oshitaka. 2020a. “ Anti - locality and subject extraction. ” Glossa 5 (1): 1–38. Erlewine, Michael Y oshitaka. 2020b. M andarin shì clefts and the syntax o f discourse congruence. Ernst, Thomas, and Cheng chi W ang. 1995. “Object pr eposing in M andarin Chinese. ” J ournal o f East Asian Linguistics 4 (3): 235–260. F anselo w , Gisbert. 2002. “ Against r emnant VP - mo v ement. ” In Dimensions o f movement: Fr om f ea- tur es to r emnants, edited b y Artemis Alexiadou, Elena Anagnostopoulou, S jef Barbiers, and H ans- M artin Gärtner , 91–125. 1. Amster dam: J ohn Benjamins. F anselo w , Gisbert. 2003. “M ünchhausen-sty le head mo v ement and the anal y sis of V erb-Second. ” In S yntax at sunset 3: H ead movement and syntactic theory, edited b y Anoop K M ahajan, 40–76. Los Angeles and P ostdam: U CLA / U niv ersita ̈t P otsdam W orking P apers in Linguistics. F anselo w , Gisbert, and Damir Ća v ar . 2002. “Distributed Deletion. ” In Theor etical Appr oaches to U ni- versals, edited b y Artemis Alexiadou, 65–107. J ohn Benjamins Publishing Compan y . 251 REFEREN CES F iantis, Dian. 1999. “ Against ‘Long H ead M o v ement’: lexical insertion and the Bulgarian auxiliary ‘B E’ . ” In T opics in south Slavic syntac and semantics, edited b y Mila Dimitr o v a- V ulchano v a and Lars H ellan, 91–124. Amster dam: J ohn Benjamins. F iengo, R obert. 1977. “On T race Theory . ” Linguistic I nquiry 8:35–61. F o x, Dann y . 2000. Economy and semantic interpr etation. 215. Cambridge, M assachusetts: MIT Pr ess. F o x, Dann y , and Da vid P esetsky . 2005. “C y clic Linearization of syntactic structur e. ” Theor etical Lin- guistics 31 (1-2): 1–46. F ukui, N aoki, and Y uji T akano. 1998. “S ymmetry in S yntax: M erge and Demerge. ” J ournal o f East Asian Linguistics 7 (1): 27–86. F unak oshi, K enshi. 2012. “On H eadless XP - M o v ement / Ellipsis On H eadless XP - M o v ement / Ellip- sis. ” 43 (4): 519–562. F unak oshi, K enshi. 2014. “S yntactic head mo v ement and its consequences. ” PhD diss., U niv ersity of M ary land, College P ark. F unak oshi, K enshi. 2019. “V erb- raising and VP - fr onting in J apanese. ” The Linguistic R eview 37 (1): 117– 146. F ung, R o xana S uk - Y ee. 2000. “F inal particles in S tandar d Cantonese: semantic extension and prag - matic inf er ence. ” PhD diss., The O hio S tate U niv ersity . Gallego, Ángel. 2010. Phase Theory. Amster dam: J ohn Benjamins. Georgi, Dor een, and Ger eon M üller . 2010. “N oun- Phrase S tructur e b y R epr ojection. ” S yntax 13 (1): 1–36. 252 REFEREN CES Gergel, R emus. 2009. M odality and Ellipsis. Diachr onic and synchr onic evidence. Berlin: M outon. Giannakidou, Anastasia, and Lisa Lai Shen Cheng. 2006. “(In )definiteness, polarity , and the r ole of w h- morp hology in fr ee choice. ” J ournal o f Semantics 23 (2): 135–183. Giorgi, Alessandra, and Giuseppe Longobar di. 1991. The S yntax o f N oun Phr ases. Cambridge: Cam- bridge U niv ersity Pr ess. Gribano v a, V era. 2017. “H ead mo v ement and ellipsis in the expr ession of R ussian polarity f ocus. ” N at- ur al Language and Linguistic Theory 35 (4): 1079–1121. Gr oat, Erich, and J ohn O’N eil. 1996. “Spell -Out at the LF Interface. ” In Minimal I deas, edited b y W erner Abraham, Samuel Da vid Epstein, H óskuldur Thráinsson, and C J an- W outer Z w art, 113–139. Amster dam: J ohn Benjamins Publishing Compan y . Gr ohmann, Kleanthes K. 2003. “S uccessiv e C y clicity U nder (Anti -)Local Considerations. ” S yntax 6 (3): 260–312. H ale, K en, and Samuel J a y K ey ser . 2002. Pr olegomenon to a theory o f argument structur e. Cambridge, M assachusetts: MIT Pr ess. H all, Da vid. 2015. “Spelling Out the N oun Phrase: Interpr etation, W or d Or der , and the Pr oblem of M eaningless M o v ement. ” PhD diss., Queen M ary , U niv ersity of London. H an, Chung - h y e, J effr ey Lidz, and J ulien M usolino. 2007. “V - Raising and Grammar Competition in K or ean: E vidence fr om N egation and Quantifier Scope. ” Linguistic I nquiry 38 (1): 1–47. H arbour , Daniel. 2008. “Discontinuous Agr eement and the S yntax - M orp hology Interface. ” In Phi The - ory, edited b y Daniel H arbour , Da vid A dger , and S usana Béjar , 185–220. O xf or d: O xf or d U ni - v ersity Pr ess. 253 REFEREN CES H arizano v , Boris. 2019. “H ead mo v ement to specifier positions. ” Glossa: a journal o f gener al linguistics 4 (1): 140. 1–36. H arizano v , Boris, and V era Gribano v a. 2019. “W hither head mo v ement?” N atur al Language and Lin- guistic Theory 37 (2 ): 461–522. H arley , H eidi. 2004. “W anting, H a ving, and Getting: A N ote on F odor and Lepor e 1998. ” Linguistic I nquiry 35 (2): 255–267. H arley , H eidi. 2013. “Diagnosing H ead M o v ement. ” In Diagnosing S yntax, edited b y Lisa Lai -Shen Cheng and N orbert Corv er , 112–120. O xf or d: O xf or d U niv ersity Pr ess. H artman, J er em y . 2011. “The S emantic U nif ormity of T races: E vidence fr om Ellipsis P arallelism. ” Lin- guistic I nquiry 42 (3): 367–388 . H eim, Ir ene. 1982. “The Semantics of Definite and Indefinite N oun Phrases. ” PhD diss., U niv ersity of M assachusetts, Amherst. H eim, Ir ene, and Angelika Kratzer . 1998. Semantics in gener ative gr ammar . M alden, MA: B lackw ell. H ein, J ohannes. 2018. “V erbal F r onting: T ypology and theory . ” PhD diss., U niv ersität Leipzig. Hinterhölzl, R oland. 2002. “R emnant mo v ement and partial deletion. ” In Dimensions o f M ovement: Fr om F eatur es to R emnants, edited b y Artemis Alexiadou, Elena Anagnostopoulou, S jef Barbiers, and H ans- M artin Gaertner , 127–150. J ohn Benjamins. Hiraiw a, K en. 2002. “Pr edicate clefts in Bùlì: categories and p hases. ” Linguistic Analysis 32 (3-4): 544– 583. H olmberg, Anders. 2000. “Scandina vian S ty listic F r onting: ho w an y category can become an exp le- tiv e. ” Linguistic I nquiry 31 (3): 445–483. 254 REFEREN CES H olmberg, Anders. 2016. The syntax o f yes and no. Cambridge: Cambridge U niv ersity Pr ess. H omer , V incent. 2015. “N eg - raising and positiv e polarity: The view fr om modals. ” Semantics and Pr ag- matics 8 (4): 1–88. H orv ath, J ulia. 1986. FOCUS in the Theory o f Gr ammar and the S yntax o f H ungarian. Dor dr echt, H ol - land: F oris Publilcations. H sieh, F eng - fan, and Rint S ybesma. 2011. “On the Linearization of Chinese Sentence. ” In K o r ean J our - nal o f Chinese Language and Liter atur e, 53–90. 1. H su, B rian. 2021. “Coalescence: a unification of bundling operations in syntax. ” Linguistic I nquiry 52 (1): 39–87. H su, Y u - yin. 2016. “Sentence- Initial M odals as F ocus Operators at CP in Chinese. ” In Pr oceedings o f the 51th Annual M eeting o f Chicago Linguistics Society, edited b y Ksenia Ersho v a, J oshua F alk, and J effr ey Geiger , 257–268. Chicago Linguistic Society . H su, Y u - yin. 2019. “M arking Pr opositional F ocus: A F unction of Pr e-S ubject M odals. ” In I ndian U ni- versity Linguistics Club W orking P apers, edited b y Kaitl yn Lee- Legg and W amsley J ames, 20–42. 2. H uang, C.- T . J ames. 1982. “Logical r elations in Chinese and the theory of grammar . ” PhD diss., M as- sachusetts Institute of T echnology . H uang, C.- T . J ames. 1987. “Existential sentences in Chinese and (in )definiteness. ” In The R epr esentation o f (I n )definiteness, edited b y Eric J R euland and Alice G B ter M eulen, 226–253. Cambridge, MA: MIT Pr ess. H uang, C.- T . J ames. 1993. “R econstruction and the structur e of VP: some theor etical consequences. ” Linguistic I nquiry 24 (1): 103–138. 255 REFEREN CES H uang, C.- T . J ames. 1994. “V erb M o v ement and Some S yntax -Semantics Mismatches in Chinese. ” Chinese Language and Linguistics 2:587–613. H uang, C.- T . J ames. 1997. “On lexical structur e and syntactic pr ojection. ” Chinese Language and Lin- guistics 3:45–89. H uang, C.- T . J ames, Y en- H ui A udr ey Li, and Y af ei Li. 2009. The syntax o f Chinese. Cambridge, MA: Cambridge U niv ersity Pr ess. H uhmarniemi, Saara. 2012. “F innish A ’ - mo v ement: Edges and I slands. ” PhD diss., U niv ersity of H elsinki. I atridou, Sabine, and H edde Zeijlstra. 2013. “N egation, P olarity , and Deontic M odals. ” Linguistic I n- quiry 44 (4): 529–568. I orio, Da vid Edy . 2015. “S ubject and object marking in Bembe. ” PhD diss., U niv ersity of N ew castle upon T yne. I srael, M. 1996. “P olarity Sensitivity as Lexical Semantics. ” Linguistics and Philosophy 19 (6): 619–666. J ack endoff, Ra y . 1997. “T wistin ’ the night a w a y . ” Language 73 (3): 534–559. J ulien, M arit. 2002. “Optional ha in S w edish and N orw egian. ” The J ournal o f Compar ative Germanic Linguistics 5 (1): 67–95. Kandybo wicz, J ason. 2008. The Gr ammar o f R epetition: N upe gr ammar at the syntax –phonolog y interf ace by. Amster dam/Philadelp hia: J ohn Benjamins Publishing Compan y . Kato, Y asuhik o. 2000. “Interpr etiv e asymmetries of negation. ” In N egation and polarity, edited b y Lau - r ence R. H orn and Y asuhik o Kato, 62–87. O xf or d: O xf or d U niv ersity Pr ess. 256 REFEREN CES Ka yne, Richar d. 1975. Fr ench S yntax: the T r ansf ormational C ycle. Cambridge, M assachusetts: MIT Pr ess. Ka yne, Richar d. 1994. The Antisymmetry o f S yntax. Cambridge, M assachusetts: MIT Pr ess. K eine, S tefan, and Rajesh B hatt. 2016. “Interpr eting v erb clusters. ” N atur al Language and Linguistic Theory 34 (4): 144 5–1492. Kim, Shin-Sook. 2002a. F ocus M atters: T wo T ypes o f I ntervention E ff ect. P aper pr esented at W CCFL 21, U C Santa Cruz. Kim, Shin-Sook. 2002b. “Interv ention Eff ects ar e F ocus. ” In J apanese/K or ean Linguistics 10, edited b y N orik o Akatsuka and S usan S trauss, 615–628. S tanf or d: CSLI Publications. Kim, Shin-Sook. 2006. “Questions, F ocus , and Interv ention Eff ects. ” In H arvar d S tudies in K or ean Linguistics XI, edited b y S usumu K uno, 520–533. 2. H arv ar d- Y enching Institute. Kim, Soo w on. 1999. “Slopp y / S trict I dentity , Empty Objects, and NP Ellipsis. ” J ournal o f East Asian Linguistics 8 (4): 255–284. Kishimoto, Hideki. 2007. “N egativ e scope and head raising in J apanese. ” Lingua 117 (1): 247–288. Kishimoto, Hideki. 2013. “V erbal comp lex f ormation and negation in J apanese. ” Lingua 135:132–154. K o, H eejeong. 2005. “S yntax of W hy-in-situ : M erge into [ Spec, CP ] in the Ov ert S yntax. ” N atur al Lan- guage and Linguistic Theory 23 (4): 867–916. K o, H eejeong. 2007. “ Asymmetries in Scrambling and C y clic Linearization. ” Linguistic I nquiry 38 (1): 49–83. K oeneman, O laf. 2000. The Flexible N atur e o f V erb M ovement. U tr echt: LO T . 257 REFEREN CES K oopman, Hilda. 1984. The S yntax o f V erbs. Dor dr echt, The N etherlands: F oris Publications. K oopman, Hilda. 1997. “U nifying Pr edicate Cleft Constructions. ” In Pr oceedings o f the 23r d Annual M eeting o f the Berkeley: Special Session on S yntax and Semantics in A frica, 71–85. K oopman, Hilda J udith, and Anna S zabolcsi. 2000. V erbal complex es. Cambridge, M assachusetts: MIT Pr ess. K otek, H adas. 2016. “On the semantics of w h-questions. ” In Pr oceedings o f Sinn und Bedeutung 20, edited b y N adine Bade, P olina Ber ezo v ska y a, and Anthea Schöller , 424–447. K otek, H adas. 2019. Composing questions. Cambridge: The MI T Pr ess. arXiv: arXiv:1011.1669v3 . Krifka, M anfr ed. 2008. “Basic notions of inf ormation structur e. ” A cta Linguistica H ungarica 55 (3-4): 243–276. Lai, J ackie Y an- ki. 2019. “P arallel cop ying in dislocation cop ying: evidence fr om Cantonese. ” J ournal o f East Asian Linguistics 3:243–277. Lambo v a, M ariana. 2004. “On T riggers of M o v ement and Eff ects at the Interfaces. ” In S tudies in Gener - ative Gr ammar , 75: T riggers, edited b y Anne B r eitbarth and H enk v an Riemsdijk, 231–258. Berlin: De Gruyter . Landau, I dan. 2006. “Chain R esolution in H ebr ew V(P )- fr onting. ” S yntax 9 (1): 32–66. Landau, I dan. 2020. “ A Scope Argument against T -to-C M o v ement in Sluicing. ” S yntax 23 (4): 375– 393. Lapointe, S tep hen G. 1980. “ A note on Akmajian, S teele, and W aso w’ s tr eatment of certain v erb com- p lement types. ” Linguistic I nquiry 11 (4): 770–787. 258 REFEREN CES Lasnik, H o w ar d, and M amoru Saito. 1984. “On the N atur e of Pr oper Go v ernment. ” Linguistic I nquiry 15 (2): 235–290. La w , Ann. 2003. “Right disloccation in Cantonese as a f ocus- marking device. ” In U niversity College London working P apers in Linguistics 15, edited b y A d N eeleman and R eik o V ermeulen, 243–275. London: U CL. La w , P aul, and J uv énal N da yiragije. 2017. “S yntactic T ense fr om a Comparativ e S yntax P erspectiv e. ” Linguistic I nquiry 48 (4): 679–696. Lechner , W infried. 1998. “Phrasal comparativ es and D P -structur e. ” In Pr oceedings o f the N orth East Linguistic Society, edited b y Pius N T amanji and Kiy omi K usumoto, 237–252. U niv ersity of T or onto: Graduate Linguistic S tudent Association. Lechner , W infried. 2007. I nterpr etive E ff ects o f H ead M ovement. A ccessed M a y 1, 2021. Lechner , W infried. 2017. I n def ense o f semantically active head movement. P apers pr esented at W ork - shop f or M artin Prinzhorn T echnical U niv ersity V ienna, N o v ember 11, 2017. Lee, T omm y T sz - Ming. 2017. “Def ocalization in Cantonese right dislocation. ” Gengo K enkyu 152:59– 87. Lee, T omm y T sz - Ming. 2020. “Def ending the N otion of Def ocus in Cantonese. ” Curr ent R esear ch in Chinese Linguistics 99 (1): 137–152. Lee, T omm y T sz - Ming. 2021. “Right dislocation of v erbs in Cantonese: A case of head mo v ement to specifier . ” In Cr ossing-over: new insights into the dialects o f Guangdong, edited b y Choi Lan T ong and I o- K ei J oaquim K uong, 104–121. M acua: H all de Cultura. 259 REFEREN CES Lee, T omm y T sz - Ming, and V ictor J unnan P an. 2021. Licensing v P movement and ellipsis in M andarin and Cantonese. P aper pr esented at The 33 r d meeting of the N orth American Conf er ence on Chi - nese Linguistics (N A CCL-33). Lema, J osé, and M aria- Luisa Riv er o. 1990. “Long H ead M o v ement: ECP v s. HM C. ” In N orth East Lin- guistics Society, 20 :333–347. Li, Charles N ., and Sandra A. Thompson. 1981. M andarin Chinese: A Functional R ef er ence Gr ammar . Berk eley: U niv ersity of Calif ornia Pr ess. Li, H aoze, and Candice Chi - H ang Cheung. 2012. “The syntactic anal y sis of f ocus interv ention eff ects in M andarin. ” Linguistic Sciences 11 (2): 113–125. Li, H aoze, and Candice Chi - H ang Cheung. 2015. “F ocus interv ention eff ects in M andarin multip le w h-questions. ” J ournal o f East Asian Linguistics 24 (4): 361–382. Li, Y af ei. 1990. “X oBinding and V erb Incorporation. ” Linguistic I nquiry 21 (3): 399–426. Li, Y en- H ui A udr ey . 1990. Or der and constituency in M andarin Chinese. Dor dr echt: Kluw er A cademic Publishers. Li, Y en- H ui A udr ey . 2005. “Shenglue yu chengf en queshi [Ellipsis and missing objects ]. ” Language Sci- ences 4 (2): 3–19. Li, Y en- H ui A udr ey . 2014. “Born empty . ” Lingua 151 :43–68. Lin, T zong - H ong J onah. 2010. “S tructur es and functional categories of M andarin sentences. ” UST W orking P apers in Linguistics 6:41–79. Lin, T zong - H ong J onah. 2011. “F initeness of Clauses and Raising of Arguments in M andarin Chinese. ” S yntax 14 (1): 48–73. 260 REFEREN CES Lin, T zong - H ong J onah. 2012. “M ultip le- modal constructions in M andarin Chinese and their finite- ness pr operties. ” J ournal o f Linguistics 48 (1): 151–186. Lin, J o- W ang, and Chih-Chen J ane T ang. 1995. “M odals as v erbs in Chinese: a GB perspectiv e. Bulletin of the Institute of History and P hilology , A cademia S inica 66: 53–105. ” Bulltetin o f the I nstitute o f History and Philolog y, A cademia Sinica 66:53–105. Linebarger , M ar cia C. 1987. “N egativ e polarity and grammatical r epr esentation. ” Linguistics and Phi- losophy 10 :325–387. Liu, Danqing. 2004. “I dentical topics: a mor e characteristic pr operty of topic pr ominent. ” J ournal o f Chinese Linguistics 32 (1): 20–64. M a hajan, Anoop K. 2003. “W or d Or der and (R emnant ) VP M o v ement. ” In W or d Or der and Scr ambling, edited b y S imin Karimi, 217–237. O xf or d: B lackw ell. M a rantz, Alec. 1997. “N o Escape fr om S yntax: Don ’t T ry M orp hological Anal y sis in the Priv acy of y our o wn Lexicon. ” In U niversity o f P ennsylvania W orking P apers in Linguistics, edited b y Alexis Dimitriadis, Laura S iegel, Clarissa S ur ek -Clark, and Alexander W illiams, 201–225. U niv ersity of P ennsy l v ania. M a rtins, Ana M aria. 2007. “Double r ealization of v erbal copies in E ur opean P ortuguese emp hatic affirmation. ” In The Copy Theory o f M ovement, edited b y N orbert Corv er and J air o N unes, 77– 118. Amster dam and Philadelp hia: J ohn Benjamins. M a ssam, Diane. 2000. “V SO and V OS: Aspects of N iuean W or d Or der . ” In The S yntax o f V erb I nitial Languages, edited b y Andr ew Carnie and Eithne Guilf o y le, 97–116. O xf or d U niv ersity Pr ess. 261 REFEREN CES M a tthew s, S tep hen, and V irginia Y ip. 1998. “V erb-cop ying Constructions in Cantonese. ” In S tudia Linguistica Serica: Pr oceedings o f the 3r d I nternational Conf er ence on Chinese Linguistics, edited b y Benjamin T’ sou, 175–189. H ong K ong: C ity U niv ersity of H ong K ong. M a tthew s, S tep hen, and V irginia Y ip. 2011. Cantonese: A Compr ehensive Gr ammar . 2nd. London: R out - ledge. M a tushansky , Ora. 2006. “H ead M o v ement in Linguistic Theory . ” Linguistic I nquiry 37 (1): 69–109. M a tyiku, Sabina M aria. 2017. “Semantic eff ects of head mo v ement : E vidence fr om negativ e auxiliary in v ersion. ” PhD diss., Y ale U niv ersity . M a y , R obert. 1977. “The Grammar of Quantification. ” PhD diss., M assachusetts Institute of T echnol - ogy . M a y , R o bert. 1985. Logical F orm: I ts S tructur e and Derivation. Cambridge, MA: MIT Pr ess. M c Closk ey , J ames. 1996. “On the Scope of V erb M o v ement in Irish. ” N atur al Language and Linguistic Theory 14 (1): 47– 104. M c Closk ey , J ames. 2016. “Interpr etation and the typology of head mo v ement : A r e-assessment. ” W ork - shop on the S tatus of H ead M o v ement in Linguistic Theory S tanf or d, September 15, 2016. Miy aga w a, Shigeru. 2001. “EP P , scrambling, and w h- in-situ. ” In K en H ale: A lif e in Language, edited b y Michael K ensto wicz, 293–338. Cambridge, MA: MIT Pr ess. M o lnárfi, László. 2002. “F ocus and antif ocus in modern Afrikaans and W est Germanic. ” Linguistics 40 (382): 1107–1160. 262 REFEREN CES M o ltmann, F riederik e, and Anna S zabolcsi. 1994. “Scope interaction with pair - list quantifiers. ” In Pr o- ceedings o f NELS 24, edited b y M er cè Gonzàlez, 381–395. GLSA, U niv ersity of M assachusetts, Amherst. M üller , Ger eon. 2004. “V erb-Second as v P - F irst. ” The J ournal o f Compar ative Germanic Linguistics 7 (3): 179–234. N c har e, A bdoula y e Laziz. 2012. “The Grammar of Shupamem. ” PhD diss., N ew Y ork U niv ersity . N evins, Andr ew , and Prana v Anand. 2003. “Some A GREEment M atters. ” In Pr oceedings o f the W est Coast Conf er ence on F ormal Linguistics 22, edited b y M T sujimura, 370–383. Somerville, M as- sachusetts: Cascadilla Pr ess. N i lsen, Øy stein. 2003. “Eliminating P ositions: S yntax and semantics of sentence modification. ” PhD diss., U tr echt U niv ersity . N i shiy ama, K unio, and E un Cho. 1998. “Pr edicate Cleft Constructions in J apanese and K or ean: The R ole of Dumm y V erbs in TP /VP Pr eposing. ” J apanese/K or ean Linguistics 7: 463–479. N unes, J air o. 1995. “The cop y theory of mo v ement and linearization of chains in the Minimalist Pr o- gram. ” PhD diss., U niv ersity of M ary land. N unes, J air o. 1998. “Bar e X -bar theory and structur es f ormed b y mo v ement. ” Linguistic I nquiry 29 (1): 160–168. N unes, J air o. 2004. Linearization o f Chains and Sidewar d M ovement. Linguistic Inquiry M onograp hs. Cambridge, M assachusetts: MIT Pr ess. N unes, J air o. 2011. “The Cop y Theory . ” In The O xf or d H andbooks in Linguistics Minimalism, edited b y Cedric Boeckx, 143–172. O xf or d: O xf or d U niv ersity Pr ess. 263 REFEREN CES N unes, J air o, and R onice M uller de Quadr os. 2006. “Dup lication of wh- elements in B razilian S ign Language. ” In Pr oceedings o f NELS 35, edited b y L. Bateman and C. U ssery , 466–477. Amherst: GLSA. O ku, Satoshi. 1998. “ A Theory of Selection and R econstruction in the Minimalist P erspectiv e. ” PhD diss., U niv ersity of Connecticut. P a n, V ictor J unnan. 2011. “ A TB-topicalization in M andarin Chinese: an Intersectiv e Operator Anal - y sis. ” Linguistic Analysis 37 (1-2): 231–272. P a n, V ictor J unnan. 2014. “W h-ex -situ in M andarin Chine se: M apping Betw een Inf ormation S truc - tur e and Sp lit CP . ” Linguistic Analysis 39 (3-4): 371–414. P a n, V ictor J unnan. 2017. “Optional pr ojections in the left -perip hery in M andarin Chinese. ” In S tudies in S yntactic Cartogr aphy, edited b y F uzhen S i, 2 16–248. Beijing: China Social Sciences Pr ess. P a n, V ictor J unnan. 2019. Ar chitectur e o f the periphery in Chinese. N ew Y ork: R outledge. P a n, V ictor J unnan. 2020. “Deriving H ead- F inal Or der in the P erip heral Domain of Chinese. ” Linguistic I nquiry: 1–34. P a ris, M arie-Claude. 1979. “Some aspects of the syntax and semantics of the ”lian.... Y e/DO U” con- struction in mandarin. ” Cahiers de Linguistique d’ Asie Orientale 5:47–70. P a ris, M arie-Claude. 1998. “F ocus operators and types of pr edication in M andarin. ” Cahiers de lin- guistique - Asie orientale 27 (2): 139–159. P a ul, W altraud. 2021. “N obody ther e? On the non-existence of nobody in M andarin Chinese and r e- lated issues. ” Canadian J ournal o f Linguistics: 1–38. 264 REFEREN CES P esetsky , Da vid. 1998. “Some Optimality Princip les of Sentence Pr onunciation. ” In I s the Best Good Enough?, edited b y Pilar Barbosa, Dann y F o x, P aul H agstr om, M artha M c Ginnis, and Da vid P e- setsky , 337–384. Cambridge, M assachusetts: MIT Pr ess. P esetsky , Da vid. 2013. Russian case morpholog y and the syntactic categories. Cambridge, MA: The MIT Pr ess. P esetsky , Da vid. 2020. “The unity of mo v ement. ” Lectur es giv en at The S t. P etersburg Institute of Lin- guistics, Cognition / Cultur e (NYI), J ul y 20-31, 2020. P esetsky , Da vid, and Esther T orr ego. 2 007. “The syntax of v aluation and the interpr etability of f ea- tur es. ” In Phr asal and Clausal Ar chitectur e: S yntactic Derivation and I nterpr etation, edited b y S imin Karimi, V ida Samiian, and W endy K W ilkins, 262–294. Amster dam: J ohn Benjamins Publishing Compan y . Platzack, Christer . 2013. “H ead mo v ement as a p honological operation. ” In Diagnosing S yntax, edited b y Lisa Lai -Shen Cheng and N orbert Corv er , 21–43. O xf or d: O xf or d U niv ersity Pr ess. P ollock, J ean- Y v es. 1989. “V erb M o v ement, U G and the S tructur e of IP . ” Linguistic I nquiry 20 (3): 365– 424. Pr eminger , Omer . 2019. “W hat the PCC tells us about “ abstract” agr eement, head mo v ement, and locality . ” Glossa: a journal o f gener al linguistics 4 (1): 1–42. Rack o w ski, Andr ea, and Lisa T ra vis. 2000. “V - initial Languages: X or XP M o v ement and A dv erbial Placement. ” In The S yntax o f V erb I nitial Languages, edited b y Andr ew Carnie and Eithne Guil - f o y le, 117–142. Ramchand, Gillian, and P eter S v enonius. 2014. “Deriving the functional hierar ch y . ” Language Sciences 46:152–174. 265 REFEREN CES Richar ds, M ar c. 2009. “Internal pair - merge: The missing mode of mo v ement. ” Catalan J ournal o f Lin- guistics 8:55–73. Richar ds, N orvin. 1997. “Competition and disjoint r ef er ence. ” Linguistic I nquiry 28 (1): 178 –187. Richar ds, N orvin. 1998. “The princip le of minimal comp liance. ” Linguistic I nquiry 29 (4): 599–629. Richar ds, N orvin. 2001. M ovement in language: inter actions and ar chitectur es. O xf or d: O xf or d U niv er - sity Pr ess. Riv er o, M aria- Luisa. 1991. “Clitic and NP Climbing in O ld Spanish. ” In Curr ent S tudies in Spanish Linguistics, edited b y H ector Campos and F ernando M artínez -Gil, 241–282. W ashington, D . C.: Georgeto wn U niv ersity Pr ess. Riv er o, M aria- Luisa. 1993. “Long head mo v ement v s. V2, and null subjects in old R omance. ” Lingua 89 (2-3): 217–245. Riv er o, M aria- Luisa. 1994. “Clause S tructur e and V - M o v ement in the Languages of the Balkans. ” N at- ur al Language and Linguistic Theory 12 (1): 63–120. Rizzi, Luigi. 1990. R elativized minimality. Cambridge, M assachusetts: MIT Pr ess. Rizzi, Luigi. 1997. “The fine structur e of the left perip hery . ” In Elements o f gr ammar , edited b y Liliane H aegeman, 281–337. Dor dr echt: Kluw er A cademic Publishers. Rizzi, Luigi. 2001. “R elativized Minimality Eff ects. ” In The handbook o f contempor ary syntactic theory, edited b y M ark Baltin and Chris Collins, 89–110. M alden, MA: B lackw ell. Rizzi, Luigi. 2004. “On the Cartograp h y of S yntactic S tructur es. ” In The S tructur e o f CP and IP , edited b y Luigi Rizzi, 3–15. O xf or d: O xf or d U niv ersity Pr ess. 266 REFEREN CES Rizzi, Luigi. 2011. “Minimality . ” In The O xf or d H andbook o f Linguistic Minimalism, edited b y Cedric Boeckx, 220–238. O xf or d: O xf or d U niv ersity Pr ess. Rizzi, Luigi, and I an R oberts. 1989. “Comp lex in v ersion in F r ench. ” Pr obus 1 (1 ): 1–30. R oberts, I an. 1991. “NP - M o v ement, Cr osso v er and Chain- F ormation. ” In R epr esentation and Deriva- tion in the Theory o f Gr ammar (22), edited b y H ubert H aider and Klaus N etter , 17–52. Dor dr echt: Kluw er A cademic Publishers. R oberts, I an. 1994. “T w o T ypes of H ead M o v ement in R omance. ” In V erb M ovement, edited b y Da vid Lightf oot and N orbert H ornstein, 207–242. Cambridge: Cambridge U niv ersity Pr ess. R oberts, I an. 2001. “H ead M o v ement. ” In The H andbook o f Contempor ary S yntactic Theory, 2nd, edited b y M ark Baltin and Chris Collins, 113–147. O xf or d: B lackw ell. R oberts, I an. 2010. A gr eement and head movement: clitics, incorpor ation, and def ective goals. Cambridge: The MIT Pr ess. R oberts, I an. 2011. “H ead M o v ement and the Minimalist Pr ogram. ” In The O xf or d H andbook o f Lin- guistic Minimalism, edited b y Cedric Boeckx, 195–219. O xf or d: O xf or d U niv ersity Pr ess. R ochemont, Michael. 1986. F ocus in gener ative gr ammar . Amster dam, The N etherlands: J ohn Ben- jamins Publishing Compan y . R ooth, M ats. 1985. “ Association with F ocus. ” PhD diss., U niv ersity of M assachusetts, Amherst. R ooth, M ats. 1992. “ A theory of f ocus interpr etation. ” N atur al Language Semantics 1 (1): 117–121. R oss, J ohn. 1967. “Constraints on v ariables in syntax. ” PhD diss., M assachusetts Institute of T echnol - ogy . 267 REFEREN CES Sag, I v an. 1976. “Deletion and logical f orm. ” PhD diss., M assachusetts Institute of T echnology . Saito, M amoru. 2003. “ A deriv ational appr oach to the interpr etation of scrambling chains. ” Lingua 113 (4-6): 481–518. Sato, Y osuk e, and M asak o M aeda. 2021. “S yntactic H ead M o v ement in J apanese: E vidence fr om V erb- Echo Answ ers and N egativ e Scope R ev ersal. ” Linguistic I nquiry, no. earl y access: 1–18. Schoorlemmer , Erik, and T anja T emmerman. 2012. “H ead mo v ement as a P F -p henomenon: E vidence fr om ident ity under ellipsis. ” In Pr oceedings o f the 29th W est Coast Conf er ence on F ormal Linguis - tics, 232–240. Somerville, M assachusetts: Cascad illa Pr ess. Shi, Dingxu. 1994. “The N atur e of Chinese W h-Questions. ” N atur al Language and Linguistic Theory 12 (2): 301–334. Shi, Dingxu, Canlong W ang, and Zhiyu Zhu. 2002. “Xianggang shumian han yu jufa bian yi: yueyu de iy ou, w en y an de bao liu ji qita [ S yntactic change on H ong K ong written Chinese: the change, f ormal sty le and others in Canonese ]. ” Applied Lin guistics 3:23–32. Shibata, Y oshiyuki. 2015. “Exp loring S yntax fr om the Interfaces. ” PhD diss., U niv ersity of Connecticut. Shimo y ama, J unk o. 2006. “Indeterminate Phrase Quantification in J apanese. ” N atur al Language Se - mantics 139–173 (14): 2. Sh yu, Shu - ing. 1995. “The S yntax of F ocus and T opic in M andarin Chinese. ” PhD diss., U niv ersity of Southern Calif ornia. Sh yu, Shu - ing. 2004. “(A)symmetries betw een M andarin Chinese lian...dou and shenzhi. ” J ournal o f Chinese Linguistics 32 (1): 81–128. Sh yu, Shu - ing. 2016. “Minimizers and ev en. ” Linguistics 54 (6): 1355–1395. 268 REFEREN CES S impson, Andr ew . 2014. “Sentence- F inal P articles. ” In The H andbook o f Chinese Linguistics, edited b y C.- T . J ames H uang, Y en- hui A udr ey Li, and Andr ew S impson, 156–179. O xf or d: J ohn W iley / Sons. S impson, Andr e w , and So y oung P ark. 2019. “S trict v s. F r ee w or d or der patterns in K or ean nominal p hrases and C y clic Linearization. ” S tudia Linguistica 73 (1): 139–174. S impson, Andr ew , and Zoe W u. 2002. “U nderstanding cy clic Spell -Out. ” In Pr oceedings o f N orth East Linguistic Society 32, edited b y M asak o Hir otani, 2 :499–518. Soh, H ooi Ling. 1998. “Object scrambling in Chinese. ” PhD diss., M assachusetts Institute of T echnol - ogy . Soh, H ooi Ling. 2005. “ W h - in-S itu in M andarin Chinese. ” Linguistic I nquiry 36 (1): 143–155. S tepano v , Arthur . 2012. “V oiding island eff ects via head mo v ement. ” Linguistic I nquiry 43 (4): 680–693. S urán yi, Balázs. 2005. “H ead mo v ement and r epr ojection. ” Annales U niversitatis Scientiarum Budapest i- nensis de R olando Eötvös N ominatae. Sectio Linguistica 26:313–342. S urán yi, Balázs. 2008. “The theory of head mo v ement and cy clic spell out. ” In Sound o f silence: Empty elements in syntax and phonolog y, edited b y J utta H artmann, V er onika H egudus, and H enk v an Riemsdijk, 293–337. Amster dam: Elsevier . S v enonius, P eter . 1994. “C-seletion as f eatur e-checking. ” S tudia Linguistica 48 (2): 133–155. S v enonius, P eter . 2016. “Spans and w or ds. ” In M orphological M etatheory, edited b y Daniel S iddiqi and H eidi H arley , 201–222. Amster dam/Philadelp hia: J ohn Benjamins Publishing Compan y . S ybesma, Rint. 1999. “The M andarin VP . ” 269 REFEREN CES S zabolcsi, Anna. 2010. Quantification. Cambridge: Cambridg e U niv ersity Pr ess. S zabolcsi, Anna. 2011. “Certain v erbs ar e syntacticall y exp licit quantifieers. ” The Baltic I nternational Y earbook o f Cognition, Logic and Communication 6:1–26. T akahashi, Daik o. 1990. “N egativ e polarity , p hrase structur e, and the ECP . ” English Linguistics 7:129– 146. T akahashi, Daik o. 2002. “Determiner raising and scope shift. ” Linguistic I nquiry 33 (4): 575–615. T akano, Y uji. 2014. “J apanese S yntax in Comparativ e P ersp. ” In J apanese S yntax in Compar ative P er - spective, edited b y M amoru Saito, 139–180. O xf or d: O xf or d U niv ersity Pr ess. T akita, K ensuk e. 2010. “C y clic Linearization and Constraints on M o v ement and Ellipsis. ” PhD diss., N anzan U niv ersity . T ang, S ze- W ing. 1998a. “On the In v erted Double Object Construction. ” In S tudies in Cantonese Lin- guistics, edited b y S tep hen M atthew s, 35–52. H ong K ong: Linguistic Societ y of H ong K ong. T ang, S ze- W ing. 1998b. “P arametrization of f eatur es in syntax. ” PhD diss., U niv ersity of Calif ornia, Irvine. T ang, S ze- W ing. 2001. “The ( non-)existence of gapping in Chinese and its imp licaitons f or the theory of gapping. ” J ournal o f East Asian Lin guistics 10 (3): 201–224 . T ang, S ze- W ing. 2002. “F ocus and dak in Cantonese. ” J ournal o f Chinese Linguistics 30 (2): 266–309. T ang, S ze- W ing. 2015. Jyutjyu jyuf aat gongji [Lectur es on Cantonese Gr ammar ]. H ong K ong: The Com- mer cial Pr ess. 270 REFEREN CES T o y oshima, T akashi. 2000. “H eading f or their o wn p laces. ” Pr oceedings o f t he 9th S tudent Conf er ence in Linguistics ( SCIL 9): 93–108. T o y oshima, T akashi. 2001. “H ead-to-spec mo v ement. ” In The minimalist par ameter: Selected papers fr om the Open Linguistics F orum, edited b y Galina M. Alexandr o v a and O lga Arnaudo v a, 115–136. Amster dam: J ohn Benjamins. T ra vis, Lisa. 1984. “P arameters and Eff ects of W or d Or der V ariation. ” PhD diss., M assachusetts Insti - tute of T echnology . T rinh, T ue. 2009. “ A constraint on cop y deletion. ” Theor etical Linguistics 35 (2-3): 183–227. T ruck enbr odt, H ubert. 2006. “On the semantic motiv ation of syntactic v erb mo v ement to C in Ger - man. ” Theor etical Linguistics 32 (3): 257–306. T sai, W ei - T ien Dy lan. 1994. “On Economizing the Theory of A - Bar Dependencies. ” PhD diss., MIT . T sai, W ei - T ien Dy lan. 2015. “On the T opograp h y of Chinese M odals. ” In Beyond Functional Sequence, edited b y U r Shlonsky , 275–294. N ew Y ork: O xf or d U niv ersity Pr ess. T sai, W ei - T ien Dy lan. 2021. “On app licativ e W h y -questions in Chinese. ” In W hy is ‘W hy’ U nique?, edited b y Gabriela Soar e, 197–218. Berlin: De Gruyter M outon. U riager eka, J uan. 1998. “ A note on ri gidity . ” In P ossessors, pr edicates and movement in the determiner phr ase, edited b y Artemis Alexiadou and Chris W ilder , 361–382. Amster dam: J ohn Benjamins Publishing Compan y . Ü r ög di, Barbara. 2006. “Pr edicate fr onting and dativ e case in H ungarian. ” A cta Linguistica H ungarica 53 (3): 291–332. 271 REFEREN CES V icente, Luis. 2007. “The S yntax of H eads and Phrases: A S tudy of V erb (Phrase ) F r onting. ” PhD diss., U niv ersiteit Leiden. v on F intel, Kai, and Ir ene H eim. 2011. I ntensional semantics ( Spring 2011 edition ). U npulished lectur e notes. W echsler , S tep hen. 1991. “V erb Second and Illocutionary F or ce. ” In Views on Phr ase S tructur e, S tudies in N atur al Language and Linguistic Theory, vol 25, edited b y Katherine Leff el and Denis Bouchar d, 177–191. Kluw er A cademic Pr ess. W ei, W ei H aley , and Y en- H ui A udr ey Li. 2018. “ A dv erbial Clauses in M andarin Chinese. ” Linguistic Analysis 1-2:163–330. W iklund, Anna- Lena. 2010. “In sear ch of the f or ce of dependent v erb second. ” N or dic J ournal o f Lin- guistics 33 (1): 81–91. W iland, Bartosz. 2008. “C ir cumstantial evidence f or syntactic head mo v ement. ” In Pr oceedings o f the 27th W est Coast Conf er ence on F ormal Linguistics, edited b y N atasha Abner and J ason B ishop, 27:440–448. Somerville, MA: Cascadilla Pr oceedings Pr oject. W ilder , Chris. 1994. “Coor dination, A TB, and Ellipsis. ” Gr oninger Arbeiten zur Germanistischen Lin- guistik 37 (1991): 291–331. W illiams, Edwin. 1978. “ A cr oss-the- Boar d R ule App lication. ” Linguistic I nquiry 9 (1): 31–43. W u, Dazhen. 2020. “ Y ueyu ’ shi -guo ’ de cihuihua ji yuji y anbian [Lexicalization and Semantic Change: Cantonese S i3 G w o3]. ” Curr ent R esear ch in Chinese Linguistics 99 (2): 359–374 . W urmbrand, S usi. 1999. “M odal v erbs must be raising v erbs. ” Pr oceedings o f the W est Coast Conf er ence on F ormal Linguistics 18: 599–612. 272 REFEREN CES Xiang, Ming. 2008. “Plurality , maximality and scalar inf er ences: A case study of M andarin Dou . ” J our - nal o f East Asian Linguistics 17 (3): 227–245. Y ang, Barry Chung - Y u. 2008. “Interv ention eff ects and the co v ert component of grammar . ” PhD diss., N ational T sing H ua U niv ersity , H sinchu. Y ang, Barry Chung - Y u. 2012. “Interv ention eff ects and wh -construals. ” J ournal o f East Asian Linguistics 21 (1): 43–87. Y ang, Xiaolong, and Y icheng W u. 2019. “ A dynamic account of lian…dou in Chinese v erb doubling cleft construction. ” Lingua 217:24–44. Y atsushir o, Kazuk o. 2009. “The distribution of quantificational suffix es in J apanese. ” N atur al Language Semantics 17 (2): 141–173. Y ip, Ka- F ai. 2020. “S yntax -pr osody M apping of right -dislocation in Cantonese and M andarin. ” In Phonologiccal Externalization volume 5, edited b y Hisao T okizaki, 73–90. Sappor o: Sappor o U ni - v ersity . Y ip, Ka- F ai, and Comf ort Ahenk orah. 2022. N on-agr eeing r esumptive pr onouns and partial Copy Dele - tion. P aper p r esented at the 96th Annual M eeting of the LSA, W ashington, DC, J an 6-9, 2022. Y ip, Ka- F ai, and T omm y T sz - Ming Lee. 2020. “Generalized Scope Econom y . ” In Pr oceedings o f the 32 nd N orth American Conf er ence on Chinese Linguistics (N A C CL-32), edited b y Kaidi Chen, 345–360. S torrs: U niv ersity of Connecticut. Y ip, M oira. 1988. “T emp late morp hology and the dir ection of association. ” N atur al Language and Lin- guistic Theory 6 (4): 5 51–577. 273 REFEREN CES Y uan, Michelle. 2017. “M or e on U ndermerge: p hrasal and head mo v ement interaction in Kikuyu. ” In A P esky Set: P apers f or David P esetsky, edited b y Clair e H alpert, H adas K otek, and Coppe v an U rk, 543–552. Cambridge, MA: MIT W orking P apers in Linguisitcs. Zeijlstra, H edde. 2017. “T w o v arieties of K or ean. ” Pr esentation at S inn und Bedeutung 22 (Micr o v ari - ation in Semantics ) on Sept 6, 2017. Z ubizarr eta, M aría Luisa, and J ean- R oger V ergnaud. 2017. “Phrasal S tr ess and S yntax. ” In The Wi- ley Blackwell Companion to S yntax ( Second Edition ), edited b y M artin E v eraert and H enk C. v an Riemsdijk. J ohn W iley / Sons. Z w art, C J an- W outer . 2001. “Object shift with raising v erbs. ” Linguistic I nquiry 32 (3): 547–554. 274

Abstract (if available)

Abstract Displacement (of linguistic expressions) is a ubiquitous phenomenon in natural language. In the generative tradition, displacement is modelled in terms of transformation, or more precisely, movement, which establishes dependencies among syntactic constituents in a phrase structure. This thesis probes the question regarding to what extent movement theories can be unified. Specifically, I address issues surrounding the debate of the distinction between head movement and phrasal movement over the past few decades. The distinction presupposes that structural complexity of the moving element is correlated with its movement properties. The goal of this thesis is to show that this is an unwarranted assumption. Based on a number of case studies on verb displacement phenomena in Cantonese, I attempt a unified theory of movement by abandoning the head/phrase distinction in movement theories. Particularly, I show (i) that verbs in Cantonese can undergo syntactic move- ment to the peripheral position of a sentence and is subject to general locality/minimality constraints on movement, and (ii) that their movement may affect semantic interpretation, leading to discourse effects and scope effects that are commonly observed in phrasal movement. I further argue, with evidence from linearization, that head movement and phrasal movement in Cantonese are subject to the same mechanism when determining the pronunciation of the movement chains. These observations converge on the conclusion that the phrase structure status of syntactic constituents bears a minimal role in theorizing displacement phenomena in natural language. This thesis represents a minimalist pursuit of a unified theory of movement.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Silence in answers: a study of ellipsis in Hindi

PDF

Syntactic derivation and the theory of matching contextual features

PDF

Syntax-prosody interactions in the clausal domain: head movement and coalescence

PDF

Perspective in Turkish complementation

PDF

Copy theory of movement and PF conditions on spell-out

PDF

Subject and adjunct island effects through the lens of selective opacity in Bangla subjects and adjuncts

PDF

Functional categories: the syntax of DP and DegP

PDF

The case of a person: The person case constraint in German

PDF

Building phrase structure from items and contexts

PDF

Number marking and definiteness in Bangla

PDF

The morphosyntax of states: deriving aspect and event roles from argument structure

PDF

Where number lies: plural marking, numerals, and the collective-distributive distinction

PDF

A unified syntactic account of Mandarin subject nominals

PDF

Syntactic and non-syntactic factors in reflexive pronoun resolution in Mandarin Chinese

PDF

Processing the dynamicity of events in language

PDF

The grammar of individuation, number and measurement

PDF

A reduplicative analysis of sentence modal adverbs in Spanish

PDF

Building adjectival meaning without adjectives

PDF

Comparative iIlusions at the syntax-semantics interface

PDF

Towards a correlational law of language: three factors constraining judgement variation

Asset Metadata

Creator Lee, Tommy Tsz-Ming (author)

Core Title Towards the unity of movement: implications from verb movement in Cantonese

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Linguistics

Degree Conferral Date 2022-08

Publication Date 07/21/2022

Defense Date 04/14/2022

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag doubling,head movement,interpretive effects,locality,OAI-PMH Harvest,phrasal movement,syntax,the unity of movement

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Li, Yen-Hui Audrey (committee chair), Simpson, Andrew (committee chair), Goodman, Jeremy (committee member), Keine, Stefan (committee member), Pancheva, Roumyana (committee member)

Creator Email tszmingl@usc.edu,tszmingtszming@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC111373667

Unique identifier UC111373667

Legacy Identifier etd-LeeTommyTs-10885

Document Type Dissertation

Format application/pdf (imt)

Rights Lee, Tommy Tsz-Ming

Type texts

Source 20220721-usctheses-batch-958 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu