
6ae131623a21d8f5fe7c1a5058473bfe.ppt
- Количество слайдов: 42
Quantifier Raising in a Top-Down Grammar Valentina Bianchi & Cristiano Chesi University of Siena XVII Colloquium on Generative Grammar Girona, June 13, 2007
The initial assimilation • (1) QR is another instance of Move a. Mary see who b. Mary saw nobody • Motivation for LF-covert syntax • (2) Similar locality constraints (Cecchetto 2004) a. A technician will complain [if you damage every plane]. ( > , * > ) plane b. *What will a technician complain [if you damage t]? (3) a. Which movie did you see t? a'. *Which movie did you see t and appreciate ‘‘The House of Mirth’’? b. A (different) student likes every professor. ( > , > ) professor b'. A (#different) student likes every professor and hates the dean. ( > , * > ) Who did Mary see t? Nobody Mary saw t. Bianchi & Chesi – QR in a Top-Down Grammar
Move, as of June 2007 a) Agree as a sub-operation of Move is feature-driven b) Phase Impenetrability Condition Move is successive-cyclic c) The “strong” topmost occurrence is spelled out Move is overt and (generally) leftward Bianchi & Chesi – QR in a Top-Down Grammar
QR, as of June 2007 a) Standard QR is not feature-driven and doesn’t target a specific position: free adjunction QR = A-movement (Hornstein, 1995) QR targets fixed positions (Beghelli & Stowell, 1997) Exception: negative and focussed phrases (Longobardi 1991, Kayne 1981)
QR, as of June 2007 b) Standard QR is not successive-cyclic; it is clause-bound successive-cyclic (tensed clause boundary). (4) Someone expected [CP that every Republican would win]. ( > ; * > ) (Exception: indefinites, Reinhart 1997) ≠ Cecchetto (2004): QR obeys the PIC. “Cyclic” steps are possible if they are semantically motivated. Bianchi & Chesi – QR in a Top-Down Grammar
QR, as of June 2007 c 1) Standard QR is covert (but see Szabolsci 1997, Kayne 1998). With cyclic Transfer, this no longer follows from the architecture of the grammar. c 2) Fox & Nissenbaum (1999), Fox (2002): QR must be rightward. (5) a. We [[saw a painting] yesterday] b. We [[[saw a painting] yesterday] <a painting>] c. We [[[saw a painting] yesterday] <a painting> by John] Bianchi & Chesi – QR in a Top-Down Grammar
Tacking stock Movement Feature-driven Successive cyclic Overt, leftward QR • QR is not a well-behaved instance of Move (e. g. Beghelli & Stowell: feature-driven, but not cyclic; Cecchetto: PIC-compliant, but not feature-driven). • The covert nature of QR no longer follows from the architecture of the grammar. Bianchi & Chesi – QR in a Top-Down Grammar
Reversing the perspective Main claim: The exceptional properties of QR follow if we assume a topdown, left-to-right computation divided in phases (Phillips 1996, Chesi 2004, Bianchi & Chesi 2006) In a nutshell: • remove the QP since not LF-interpretable • re-merge it in a position when it can take an adequate argument/nuclear scope Bianchi & Chesi – QR in a Top-Down Grammar
Formalizing a minimalist grammar 1. Feature Structures (lexicon + parameterization) 2. Universals (structural constraints + economy conditions) 3. Structure Building (merge, move, phase) Operations Bianchi & Chesi – QR in a Top-Down Grammar
1 – Feature Structures base = {N, V} Only TWO main categories: Nouns and Verbs (e. g. [V give]) select = {base licensors} Lexical selection (e. g. [=DP =PP V give]) Ordered sets of functional features (e. g. [=DP =PP. . . +Mood. . . +T. . . V give]; DP [ +D N ]; PP [ +K +D N ] ) Bianchi & Chesi – QR in a Top-Down Grammar
2 – Universals Linearization Principle (inspired by LCA, Kayne 1994) if A dominates B, then either a. A precedes B if B is a complement of A (that is, A selects B), or b. B precedes A if B is in a functional projection of A B dominance: B>A B>C B A B C + time precedence: <A, B, C> Bianchi & Chesi – QR in a Top-Down Grammar
3 - Structure Building Operations a. MERGE (merge right, Phillips 1996) Lexicon: {[=DP =PP V gives], [+K (N) to], [+D N John], [+D N children], [+D N candies]} 1. merge ([=DP =PP V gives], [+D N John]) V V gives N John
3 - Structure Building Operations a. MERGE (merge right, Phillips 1996) Lexicon: {[=DP =PP V gives], [+K (N) to], [+D N John], [+D N children], [+D N candies]} 1. merge ([=DP =PP V gives], [+D N John]) V V gives V N John V <gives>
3 - Structure Building Operations a. MERGE (merge right, Phillips 1996) Lexicon: {[=DP =PP V gives], [+K (N) to], [+D N John], [+D N children], [+D N candies]} 1. merge ([=DP =PP V gives], [+D N John]) 2. merge ([=DP =PP V gives], [+D N candies]) V V gives V N John V V <gives> V N candies V <gives>
3 - Structure Building Operations a. MERGE (merge right, Phillips 1996) Lexicon: {[=DP =PP V gives], [+K (N) to], [+D N John], [+D N children], [+D N candies]} 1. merge ([=DP =PP V gives], [+D N John]) 2. merge ([=DP =PP V gives], [+D N candies]) 3. merge ([=DP =PP V gives], [+K +D N to children]) V V gives V N John V V <gives> V N candies V <gives> V N to children
3 - Structure Building Operations c. MOVE b. PHASE (PROJECTION) Linearization Principle (inspired by Kayne’s LCA) if A immediately dominates B, then either a. <A, B> if A selects B as an argument, or b. <B, A> if B is in a functional specification of A V Force. . . (left periphery) V e. g. “the boy kissed the girl” V V Mood V . . . Functional Sequence [+T kiss] Asp V the boy [=s =o kiss] V (licensor features) head kissed V <the boy> Selected Phase(s) Phase selection requirement: phases must be properly (licensed/)selected [=o kiss] the girl (select features) M(ove)-Buffer [=s =o kiss] the boy M-Buffer Success condition: M-buffer(s) must be empty at the end of the computation
Successive Cyclic A'-movement who = 1 st Nested Phase (DP) Who Lic. you = 2 nd Nested Phase (DP) do Matrix Phase (CP) you Sel. that = Selected Phase (CP) believe you who M(ove)-Buffer (Matrix Phase, CP) <you> <who> that John who admires (6) Whoi do you believe [twho that John admires twho]? <who> Bianchi & Chesi – QR in a Top-Down Grammar
To summarize 1. Every computation is a top-down process divided in phases Each lexical phase head licenses a left-hand functional domain and some right-hand selected positions. 2. A phase n gets closed when all the functional and selectional specifications of its head are satisfied. Any internal constituent will be a computationally nested phase 3. The Move operation stores an unselected element found before (i. e. on the left of) the head position in the local Mbuffer of the current phase, and discharges it in a selected position if possible; if not, when the phase is closed the content of the memory buffer is inherited by that of the lowest selected phase (the sequential phase, Chesi 2004). Bianchi & Chesi – QR in a Top-Down Grammar
A fundamental asymmetry Overt movement: the system first computes the displaced occurrence in a functionally licensed (criterial) position, stores the element in a M(ove)-memory buffer, and then buffer looks for a selected position where the element can be remerged. 1 What P 1 P 2 M 1 P 2 2 P 3 did P 3 John 3 <P 3> buy P 4 t. P 3 4 <P 2> P 5 Bianchi & Chesi – QR in a Top-Down Grammar
A fundamental asymmetry Quantifier Raising: the system computes the QP in an argument position which is PF-interpretable but not LFinterpretable, stores the QP in a Q(uantifier)-memory buffer, and re-merges it at the point where it can be properly buffer interpreted (i. e. , at the end of the phase). 1 Mary P 1 P 2 M 1 2 Q 1 3 t. P 2 every book P 2 gave P 3 P 4 4 P 5 to Sue <P 4> P 6 Bianchi & Chesi – QR in a Top-Down Grammar
An implementation of QR a. Compute a QP and spell it out in the selected (or functionally licensed) position within phase n. b. Insert the QP in the Q-buffer of phase n with index i (QPi) c. Insert a variable with index i in the selected position. d. At the end of the computation of phase n, retrieve QPi from the Q-buffer of n and attach it to the structure built in phase n. e. Success Condition: at the end of any phase n, the Q-buffer of n must be empty. Bianchi & Chesi – QR in a Top-Down Grammar
QR – sample derivation Fuctional projections (7) Mary gave every book to Sue Mary VP shells gave V <Mary> every ibook x Mary <every booki> to John M(ove)-Buffer Q(uantifier)-Buffer every booki Bianchi & Chesi – QR in a Top-Down Grammar
Main Consequences 1. The re-merge position is (as usually) covert 2. The re-merge position of QR follows the computation of the selected position: “rightward” movement 3. The clause-boundedness of QR is a “right roof” effect, corresponding to a final phase boundary. Bianchi & Chesi – QR in a Top-Down Grammar
1. Covertness • The position computed first is “PF-interpretable” (criterial or argumental position) and the QP phase is spelled out there, before storage in the Q-buffer • Remerge positions are generally unpronounced (Chesi 2004) • It is possible to implement Late Merge à la Fox & Nissenbaum (1999) 1 P 1 We P 2 Mary M 1 P 2 saw P 3 Q 1 2 t. P 2 P 4 3 a painting 4 P 5 yesterday <P 4> P 6 by John Bianchi & Chesi – QR in a Top-Down Grammar
2. Rightward orientation • The first position of the QP dependency is selected or functionally licensed. • “Rightward” movement: the re-merge position of QR follows the computation of the selected position. • The remerge position implements inverse selection: the structure previously computed in the current phase is the argument of the QP function. Bianchi & Chesi – QR in a Top-Down Grammar
3. Clause-boundedness (8) a. What did a technician say [t'' that John t' inspected t] ? b. A technician said [that John inspected every plane] plane (∃>∀; *∀>∃) c. * [every plane] a technician said [t'' that John t' inspected t] plane (cf. Cecchetto 2004: 345 ) QR is not successive cyclic: ® Why no attraction by the edge feature EF? ® Why no one-step Form Chain? Bianchi & Chesi – QR in a Top-Down Grammar
3. Clause-boundedness The clause-boundedness of QR is a right roof effect. The QP is stored in the Q-buffer of the current phase n: a) It takes scope over all the phases nested in n, by rightward attachment; b) It cannot take scope over any superordinate phase, because this would require either non-local retrieval, or supercopying from the Q-buffer of the current phase into the Qbuffer of a superordinate phase. Bianchi & Chesi – QR in a Top-Down Grammar
3. Clause-boundedness super-copying M 1 P 2 Q 1 P 2 1 A technician i said t. P 2 5 Q 4 2 P 3 P 2 3 P 4 non-local retrieval P 5 4 <P 2>i P 5 that J. inspected every plane k <P 5>k Bianchi & Chesi – QR in a Top-Down Grammar
Further Consequences 1. Scope ambiguities Surface Scope Preference 2. Pronominal Binding Leftness Condition Bianchi & Chesi – QR in a Top-Down Grammar
1. Scope ambiguities Default derivation: Surface Scope Every boy x invited 1 twox 2 girls <every boy 1> <two girls 2> Q-Buffer two girls 2 every boy 1 (9) Every boy invited two girls ( x 2. x 1 T invited x 2 ) every boy ( x 1. two girls ( x 2. x 1 T invited x 2 ))
1. Scope ambiguities Reordering in Q-buffer: Inverse Scope Every boy x invited 1 twox 2 girls <two girls 2> <every boy 1> Q-Buffer two girls 2 every boy 1 (9) Every boy invited two girls every boy ( x 1 T invited x 2 ) two girls ( x 2. every boy ( x 1 T invited x 2 ))
1. Scope ambiguities Reordering in Q-buffer: Inverse Scope Every boy x invited 1 twox 2 girls <two girls 2> <every boy 1> G/CQPs two girls 2 DQPs every boy 1 Q-Buffer two girls ( x 2. every boy ( x 1 T invited x 2 ))
2. Pronominal Binding Implementation of A-binding (Bianchi 2007 a, based on Schlenker 2005): • When an R-expression is processed, its referent is stored (step 2) in a phase-local R(eferential)-buffer (≠ M-buffer & Q-buffer: no discharge/remerge); Q-buffer • Nested and selected phases inherit the R-buffer of the containing phase (step 4) • The bound pronoun retrieves the referent (via a negative index) from within the Rbuffer (step 5, and moves it to the last position of the R-buffer, where it is used to R-buffer evaluate the truth conditions) M 1 P 1 1 John P 2 John R 1 P 2 2 loves P 3 4 R 4 3 <P 2> P 2 5 P 4 his-1 wife Bianchi & Chesi – QR in a Top-Down Grammar
2. Pronominal Binding When the pronoun is bound by a QP: • After QR (step 1), the bound variable is stored in the local R-buffer (step 3); • The pronoun retrieves the variable from the R-buffer in the usual way (step 6) Q 1 P 1 M 1 P 2 1 Every man P 2 Every man xi R 1 7 xi 5 2 3 loves R 4 4 P 3 xi xi 6 P 4 his-1 wife <every mani> Bianchi & Chesi – QR in a Top-Down Grammar
2. Pronominal Binding This mechanism immediately derives the Leftness Condition: (10) *His wife loves every man. • The pronoun can retrieve the Q-bound variable from the R-buffer only after the QP has been processed and the variable has been inserted by the QR operation. • Therefore, the processing of the QP must precede the processing of the bound pronoun. (Cf. Schlenker 2005, Shan & Barker 2006). Q 1 R 1 P 1 Every wife P 2 His man loves. . . Bianchi & Chesi – QR in a Top-Down Grammar
Conclusions Advantages of a top-down derivation: • Covertness and rightward orientation of QR • Rightward attachment as inverse selection • Right roof constraint (i. e. , limitation to the immediately containing phase) • Preference for surface scope (last in, first out retrieval strategy) • Leftness Condition on Q-binding of pronouns What remains of the initial assimilation of QR to overt instances of Move? Move • Storage mechanism, with phase-local stores (but Q-buffer instead of Mbuffer) buffer • Emptiness condition (the stored elements must be “discharged” from the store by the end of the phase computation) Bianchi & Chesi – QR in a Top-Down Grammar
Further consequences (in progress) 1. Inverse Linking 2. Lower Scope (v. P scope) w. r. t. negation and modals 3. Economy of Scope (Fox 2000) 4. Wh-/QP Scope interactions (Chierchia 1993, Beghelli & Stowell 1997) 5. Cyclic QR (Cecchetto 2004) 6. Semantic Nesting Bianchi & Chesi – QR in a Top-Down Grammar
1. Inverse Linking (11) [c. QP One apple in [i. QP every basket]] is rotten basket How to obtain wide scope of the i(nternal)QP over the c(ontaining)QP? a. Extraction of i. QP from c. QP (cf. Sauerland 2005) b. Adjunction of i. QP to c. QP (cf. Büring 2004) 3 P 1 Q 1 4 P 2 1 One apple P 2 One apple Q 2 P 3 in P 3 every basket 2 <P 3 i> is rotten <P 2 j> Bianchi & Chesi – QR in a Top-Down Grammar
1. Inverse Linking A pronoun in the matrix phase apparently bound by the i. QP must be an E-type pronoun, à la Büring (2004): (12) [c. QP Somebody from [i. QP every city]k ]i hates itsk climate city Somebody from every city hates [the city they are from]’s climate Problem: how to obtain internal scope of the i. QP? (Cf. Heim & Kratzer 1998, 221 ff. ) This may follow if the PP can be an independent phase with its own Qbuffer (i. e. , akin to a reduced relative). Bianchi & Chesi – QR in a Top-Down Grammar
2. VP-scope (13) Al didn’t attend more than two meetings (Heim & Kratzer 1998: 218) ( QP) (QP > ): a) ( QP) the maximum number of meetings that Al attended is two b) (QP > ) There are more that two meetings such that Al did not attend them Our top-down system doesn’t have a v. P phase with a Q-buffer lying in the scope of negation (cp. Fox’s v. P scope). The matrix phase Q-buffer will have scope over negation. Assume that negation too is stored in the Q-buffer, so that it can take either relative scope w. r. t. the QP. This assumption is also required to account for Quantifier Lowering of a subject QP into the scope of negation (cp. Fox’s lowering to the v. P-trace position): (14) Every arrow didn’t hit the target Bianchi & Chesi – QR in a Top-Down Grammar
3. Economy of scope (Fox 2000) (14) • • • a. A boy admires every teacher. ( > ), ( > ) teacher b. A boy admires every teacher. Mary does, too. (* > ), ( > ) teacher In order to have scope reversal in the first conjunct of (b), the QPs in the Q -buffer must be rearranged No rearrangement of the Q-buffer is required in the second conjunct, because the subject is non-quantificational therefore, the two conjuncts are not semantically parallel. Does the linear position of the scopally uninformative conjunct matter? Probably not: (14) c. Yesterday, a guard stood in front of this church, church and a policeman did, in front of every mosque (# > ), (* > ) Bianchi & Chesi – QR in a Top-Down Grammar
Selected references Beghelli, Filippo and Tim Stowell. 1997. Distributivity and negation. In Ways of scope taking, ed. Anna Szabolcsi, 77 -109. Dordrecht: Kluwer. Bianchi, Valentina. 2001. Antisymmetry and the Leftness Condition: Leftness as anti-c-command. Studia Linguistica 55, 1 -38. Bianchi, Valentina. 2007. Non-redundancy and backward anaphora. XXX Glow Colloquium, Tromsoe. Bianchi, Valentina. & Cristiano Chesi. 2006. Phases, left branch islands, and computational nesting. U. Penn Working Papers in Linguistics 12. 1, 15 -28. Büring, Daniel. 2004. Crossover situations. Natural Language Semantics 12, 23 -62. Cecchetto, Carlo. 2004. Explaining the locality conditions of QR: Consequences for theory of phases. Natural Language Semantics 12, 345 -397. Chesi, Cristiano. 2004. Phases and Cartography in Linguistic Computation. Doct diss. , University of Siena. Fox, Danny. 2000. Economy and semantic interpretation. Cambridge, Mass. : MIT Press. Fox , Danny & Nissenbaum, Jon. 1999. Extraposition and scope: A case for overt QR. Proceedings of WCCFL 18, 132 -144. Kayne, Richard S. 1998. Overt vs. covert movement. Syntax 1: 128 -191. Reinhart, Tanya. 1983. Anaphora and Semantic Interpretation. Chicago: The University of Chicago Press. Reinhart, Tanya. 1997. Quantifier scope: how labor is divided between QR and choice functions. Linguistics and Philosophy 20, 335 -397. Sauerland, Uli. 2005. DP is not a scope island. Linguistic Inquiry 36, 303 -314. Schlenker, Philippe. 2005. Non-redundancy: towards a semantic reinterpret-ation of binding theory. Natural Language Semantics 13, 1 -92. Shan, C. & Chris Barker. 2006. Explaining crossover and superiority as left-to-right evaluation. Linguistics and Philosophy 29, 91 -134. Bianchi & Chesi – QR in a Top-Down Grammar
6ae131623a21d8f5fe7c1a5058473bfe.ppt