<<

Universität Leipzig November òýÔÀ

A crossover puzzle in Hindi scrambling

Stefan Keine University of Southern California (presenting joint work with Rajesh Bhatt)

Ô Introduction Condition C, i.e., if it does not amnesty a Condition C violation in the base position (see Lebeaux ÔÀ——, òýýý, Chomsky ÔÀÀç, Sauerland ÔÀÀ—, Fox ÔÀÀÀ, • Background and terminology Takahashi & Hulsey òýýÀ). Ô. Inverse linking ( ) a. *He thinks [John’s mother] is intelligent. A possessor may bind a c-commanded by its host DP (May ÔÀÞÞ, Ô Ô b. * mother thinks is intelligent. Higginbotham ÔÀ—ý, Reinhart ÔÀ—ç). We will refer to this phenomena as [John’sÔ ]ò heÔ ò inverse linking (see May & Bale òýýâ for an overview). • Movement-type asymmetries (Ô) a. [Everyone’sÔ mother] thinks heÔ is a genius. English A- and A-movement dišer w.r.t. several of these properties: b. [No one’s mother-in-law] fully approves of her . Ô Ô (â) Weak crossover (WCO) ò. Crossover a. A-movement Crossover arises if a DP moves over a pronoun that it binds and the resulting Every boyÔ seems to hisÔ mother [ Ô to be intelligent ] structure is ungrammatical (Postal ÔÀÞÔ, Wasow ÔÀÞò). b. A-movement (ò) Strong crossover (SCO): pronoun c-commands trace *Which boyÔ did hisÔ mother say [ Ô is intelligent ]? a. *DPÔ ... pronÔ ... tÔ (Þ) Secondary strong crossover (SSCO) b. *WhoÔ does sheÔ like Ô? a. A-movement (ç) Weak crossover (WCO): pronoun does not c-command trace [ Every boy’sÔ mother ]ò seems to himÔ [ ò to be intelligent ] a. *DP ... [ ... pron ...] ... t Ô DP Ô Ô b. A-movement b. *WhoÔ does [herÔ mother] like Ô? *[ WhoseÔ mother ]ò does heÔ think [ ò is intelligent ]? ç. Secondary strong crossover (SSCO) SSCO refers to ungrammatical congurations that involve strong crossover (—) Condition C connectivity with possessors under inverse linking (van Riemsdijk & Williams ÔÀ—Ô, Postal ÔÀÀç, Sar a. A-movement

ÔÀÀÀ). [ John’sÔ mother ]ò seems to himÔ [ ò to be intelligent ] b. A-movement (¥) *[WhoseÔ mother]ò do you think sheÔ likes ò? *[ John’sÔ mother ]ò heÔ thinks [ ò is intelligent ] ¥. Condition C connectivity A movement step displays Condition C connectivity if it reconstructs for

Ô • Goal for today ò A scrambling puzzle We examine and document the apparently paradoxical behavior of local scram- bling in Hindi w.r.t. these properties. • Goal We investigate the crossover properties of local (i.e., clausebounded) scrambling ⪧ Scrambling is not subject to WCO → patterns like English A-movement in Hindi (which we will simply refer to as “scrambling” in what follows). ⪧ Scrambling displays SSCO and Condition C connectivity → patterns like English A-movement • Weak crossover A standard generalization about local scrambling in Hindi is that it is not subject • Proposal to WCO (Mahajan ÔÀÀý, Gurtu ÔÀÀò, et seq.); see (Ôý). In this respect, scrambling is apparently paradoxical behavior indicates that WCO on the one hand and clearly behaves like English A-movement (âa). SSCO and Condition C connectivity on the other have distinct sources. (Ôý) Local scrambling is not subject to WCO Ô. WCO is conditioned by the A- or A-nature of the landing site. a. behin-ne da˜at˜ aa [ us-kiiÔ~*ò ][ har lar.ke-koò ] . . ò. SSCO and Condition C connectivity are conditioned by the representation s/he-u• sister-u§ every boy-Zhh scolded of the moved DP in its launching site, specically whether the launching ‘Her/his sister scolded every boy .’ Ô~*ò ò site has received case or not. b. behin-ne da˜at˜ aa [ har lar.ke-ko ]Ô [ us-kiiÔ ] Ô . . • English A-movement: every boy-Zhh s/he-u• sister-u§ scolded movement of a non-case-marked DP to an A-position ‘For every boy x, x’s sister scolded x.’ • English A-movement: • movement of a case-marked DP to an A-position Inverse linking Hindi allows for inverse linking: a possessor inside a DP may bind a pronoun c-commanded by the container DP. (À) (Hindi) scrambling involves movement of an already case-marked DP to an A-position. (ÔÔ) by possessor behin-ne da˜at˜ aa a. Because the movement targets an A-position, it is resilient to WCO. [ har lar.ke-kiiÔ ] us-koÔ . . b. Because the movement applies to a case-marked DP, it is subject to every boy-u• sister-u§ he-Zhh scolded SSCO and Condition C connectivity. ‘For every boy x, x’s sister scolded x.’

• Incidentally, such inverse linking is possible even if the possessor demonstrably • SSCO, Condition C, and case cannot move out of the container DP. While Hindi in principle allows possessor To derive the connection between SSCO, Condition C connectivity, and case, extraction, such extraction is not possible out of ergative subjects (Ôò) or ko- we adopt Takahashi & Hulsey’s(òýýÀ) Wholesale Late Merge account, which is marked objects (Ôç). crucially linked to case. (Ôò) No possessor extraction out of ergative DPs • As a result, (Hindi) scrambling behaves like A-movement w.r.t. the properties of a. kal [ Ram-kii behin-ne ] Anu-ko d. a˜at˜.aa the landing site (WCO) but like A-movement w.r.t. the properties of the trace yesterday Ram-u• sister-u§ Anu-Zhh scolded (SSCO and Condition C). ‘Yesterday, Ram’s sister scolded Anu.’

b. *Ram-kiiÔ kal [ Ô behin-ne ] Anu-ko d. a˜at˜.aa Ram-u• yesterday sister-u§ Anu-Zhh scolded

ò (Ôç) No possessor extraction out of ko-marked objects • Another way of framing the problem a. us-ne [ Ram-kii behin-ko ] d. a˜at˜.aa s/he-u§ Ram-u• sister-Zhh scolded Scrambling can feed inverse linking (i.e., binding by a possessor inside the scram- bled DP) only if the bound pronoun does not c-command the launching site ‘S/he scolded Ram’s sister.’ Ô ò (see (Ô¥) vs. (Ô )). No analogous restriction holds in the absence of scrambling. b. *Ram-kii us-ne [ behin-ko ] da˜at˜ aa Ô Ô . . Looks like we are dealing with a constraint on scrambling. Ram-u• s/he-u§ sister-Zhh scolded ⪧ • Condition C connectivity with possessors • Inverse linking and movement Hindi scrambling is subject to Condition C connectivity with possessors. (ÔÞa) Inverse linking is in principle possible for derived structures as well. In (ÔÔ), the demonstrates that, unsurprisingly, a coindexed subject creates a Condition C vio- object har lar.ke-kii behin-ko ‘every boy’s sister-Zhh’ scrambles over the subject, lation for a possessor R-expression inside the object. (ÔÞb) shows that scrambling and it may bind the subject-internal pronoun us-ke. of the object over the subject does not obviate this Condition C violation.

(Ô¥) Binding by possessor inside scrambled DP (ÔÞ) Scrambling does not amnesty Condition C violations [ har larke-kii behin-ko ] [ us-ke dost-ne ] da˜at˜ aa . Ô ò Ô ò . . a. *us-ne [ Sita-ke bhaaii-ko ] da˜at˜ aa every boy-u• sister-Zhh he-u• friend-u§ scolded Ô Ô . . she-u§ Sita-u• brother-Zhh scolded ‘For every boy x, x’s friend scolded x’s sister.’ ‘SheÔ scolded Sita’sÔ brother.’

• Secondary strong crossover b. *[ Sita-keÔ bhaaii-ko ]ò us-neÔ ò d. a˜at˜.aa Curiously, however, a possessor contained inside a scrambled DP may not bind a Sita-u• brother-Zhh she-u§ scolded pronoun that is crossed by movement if this pronoun c-commands the launching ‘Sita’sÔ brother, sheÔ scolded.’ site: • Summary (Ô ) No binding by possessor if pronoun c-commands trace Hindi scrambling does not align with English A- or A-movement with respect * behin-ko da˜at˜ aa [ har lar.ke-kiiÔ ]ò us-neÔ ò . . to the above properties. every boy-u• sister-Zhh he-u§ scolded ‘For every boy x, x scolded x’s sister.’ (ԗ) English Hindi English A-movement scrambling A-movement • (Ô ) constitues a SSCO conguration similar to (Þb). e ungrammaticality WCO N N Y of (Ô ) thus indicates that Hindi scrambling is subject to SSCO, like English A-movement. SSCO N Y Y possessor Condition C NY Y • e puzzle connectivity Ô. Scrambling feeds variable binding (see (Ôý) and (Ô¥)); • ò. Possessors may bind outside the host DP in the absence of movement (see Questions (ÔÔ)); Ô. How does scrambling relate to the A/A-movement distinction (see, among many others, Webelhuth ÔÀ—À, Mahajan ÔÀÀý, Dayal ÔÀÀ¥)? ç. e trace in (Ô ) is not coindexed with the subject pronoun, so there is no transparent Condition B/C ešect w.r.t. the trace. ò. Does scrambling constitute a third primitive type of movement, in addition to the standard A/A-dichotomy, or can this constellation of properties be (Ôâ) *[ every boy’s sister ] ... him ... t ... Ô ò Ô ò derived from other, independent properties of scrambling? ¾ Why is (Ô ) ungrammatical?

ç ç Proposal (òÔ) (Hindi) scrambling involves movement of an already case-marked DP to an A-position. • Drawing on Takahashi & Hulsey’s(òýýÀ) Wholesale Late Merger account of Condition C connectivity, we propose that the SSCO pattern in Hindi scrambling a. Because the movement targets an A-position, it is resilient to WCO. derives from the independent fact that scrambling does not feed case assignment. b. Because the movement applies to a case-marked DP, it is subject to SSCO and Condition C connectivity. • Scrambling and case Hindi scrambling dišers from English A-movement in that it is independent of case: scrambling never ašects the case of the moving element, and it does not µ In other words, we propose that scrambling is A-movement that takes place aŸer discriminate among DPs based on their case feature (Keine òýԗ). case has been assigned to a DP and that it is this combination of properties that underlies the apparently paradoxical properties of scrambling with respect to (ÔÀ) Case connectivity: Accusative the A/A-distinction. a. Sita-ne Ram-{ko~*se~*kaa~*∅} dekhaa Sita-u§ Ram- Zhh *†•«±§ *u• * saw { ~ ~ ~ ∅} ç.Ô Weak crossover ‘Sita saw Ram.’ • b. Ram-{ko~*se~*kaa~*∅} Sita-ne dekhaa Pronominal binding Ô Ô We take for granted, as is standard, that pronominal binding is possible only Ram-{Zhh~*†•«±§~*u•~*∅} Sita-u§ saw from A-positions, but not from A-positions. ‘Sita saw Ram.’ • Possible implementations (òý) Case connectivity: Instrumental – van Riemsdijk & Williams(ÔÀ—Ô), Williams(òýýç): a. Pratap Sita-{se~*ko~*kaa~*∅} milaa hai Binding applies aŸer A-movement, but before A-movement Pratap Sita-{†•«±§~*Zhh~*u•~*∅} met Z¶ì ‘Pratap has met Sita.’ – Sauerland(ÔÀÀ—, òýý¥), Ruys(òýýý), Keine & Poole(òýԗa,b): A-movement creates λ-abstraction over e-type variables; A-movement cre- b. Sita-{se~*ko~*kaa~*∅}Ô Pratap Ô milaa hai ates λ-abstraction over higher-typed variables. Sita-{†•«±§~*Zhh~*u•~*∅} Pratap met Z¶ì ‘Pratap has met Sita.’ • Independent support I argue in Keine(òýԗ, òýÔÀ) that the type of scrambling in Hindi investigated • We conclude that scrambling does not feed case assignment, hence that scram- here lands in a TP-internal position. bling targets DPs whose case is already valued prior to scrambling (see Keine • If scrambling targets an A-position, it follows that it is not subject to WCO. òýԗ for additional arguments for this conclusion). (òò) behin-ne da˜at˜ aa µ In this respect, scrambling behaves like A-movement and unlike A-movement. [ har lar.ke-ko ]Ô [ us-kiiÔ ] Ô . . every boy-Zhh he-u• sister-u§ scolded • Proposal ‘For every boy x, x’s sister scolded x.’ We propose that the independence of scrambling and case provides the key to understanding the mixed properties of scrambling in regard to WCO, SSCO, (òç) [every boy-Zhh] λxe . [[ x’s sister-u§] scolded x ] → no WCO and Condition C connectivity. Specically, we propose (òÔ).

¥ ç.ò Strong crossover and Condition C connectivity (òÞ) Takahashi & Hulsey’s(òýýÀ) analysis of English A-movement

• We propose that Takahashi & Hulsey’s(òýýÀ) account of Condition C con- nectivity combined with (òÔ) derives the result that scrambling behaves like DP ⋯ A-movement with respect to these properties.

DP-u•Ô ... pronÔ DP → Condition C connectivity ç.ò.Ô Late Merge ... • Late Merge DP-u•Ô Lebeaux(ÔÀ——, òýýý) observes that an R-expression inside an adjunct does not give rise to Condition C connectivity under either A- or A-movement. ⪧ English A-movement Takahashi & Hulsey(òýýÀ) propose that A-movement has the option of late-merging the NP restrictor (so-called WLM). (ò¥) [ Which picture that MaryÔ took during her recent trip ]ò did sheÔ show Wholesale Late Merger, ò to Alex? – Because the launching site only contains a D head, Condition C is not violated. • Lebeaux(ÔÀ——, òýýý) proposes that the relative clause can be late-merged to the – More generally, aŸer Trace Conversion, the D head in the launching site DP aŸer this DP has undergone movement. is not dišerent from a pronoun in the relevant respects, and it is hence not subject to Condition C for principled reasons. (ò ) a. [C [sheÔ show [which picture] to Alex]]

b. [[Which picture] C [sheÔ show a[which picture]f to Alex]] (ò—) Takahashi & Hulsey’s(òýýÀ) analysis of English A-movement

c. [[Which picture] that MaryÔ took during her recent trip] C [ sheÔ show a[which picture]f to Alex]] DP ⋯ → no Condition C connectivity

µ Because the lower copy does not contain an R-expression, Condition C is re- DP-u• ... pron D spected. Ô Ô • Condition C connectivity with possessors: an asymmetry • Restricting WLM If an R-expression is embedded inside an argument or possessor, Condition C To ensure that Condition C is obviated with possessors only under A-movement, connectivity arises with A-movement, but not with A-movement. the WLM derivation in (ò—) must only be available to A-movement.

(òâ) a. A-movement • Takahashi & Hulsey(òýýÀ) do not stipulate this restriction as such, but attribute [ John’sÔ mother ]ò seems to himÔ [ ò to be intelligent ] it to case: Concretely, they assume that the NP restrictor is subject to the Case b. A-movement Filter, and as a result, the NP restrictor must be merged prior to the DP reaching a position in which its case is assigned. *[ John’sÔ mother ]ò heÔ thinks [ ò is intelligent ]

• Wholesale Late Merge (WLM) (òÀ) Restriction on WLM Takahashi & Hulsey(òýýÀ) extend Lebeaux’s(ÔÀ——, òýýý) Late Merge account WLM cannot target a D aŸer D has been assigned case. to this asymmetry: ⪧ English A-movement A-movement leaves behind a copy of the moved DP minus adjuncts. As- • English A-movement: feeds case assignment → has access to a WLM derivation. suming that possessors cannot be late-merged (Sar ÔÀÀÀ), a possessor R- • English A-movement: does not feed case assignment → WLM into an A-moved expression gives rise to Condition C connectivity. DP would violate the Case Filter.

¾ Consequence (çç) Structure for (çò): Condition C violation Only A-movement obviates Condition C violations with possessors.

ç.ò.ò to Hindi scrambling DP ⋯ • We proposed in (òÔ) that scrambling is A-movement of an already case-marked DP. Takahashi & Hulsey’s(òýýÀ) account predicts that in this case, the NP har lar.ke-kiiÔ behin-ko us-neÔ ⋯ restrictor must be merged before scrambling applies (or else the Case Filter would be violated), just as in the case of A-movement. DP d. a˜at˜.aa

⪧ Condition C connectivity behin-ko e obligatory presence of the NP restrictor then immediately results in Condi- har lar.ke-kiiÔ tion C connectivity with possessors. is between English A-movement on the one hand and English A-movement and Hindi scrambling on the other • Consequences is schematized in (çý) and (çÔ), respectively. Ô. e ungrammaticality of (çò) follows as a Condition C ešect: e launching site contains the possessor, which behaves like an R-expression. Coindexa- (çý) English A-movement: no Condition C connectivity; no (S)SCO tion with a c-commanding QP leads to ungrammaticality. [ [ DP-u•Ô ...] ... pronÔ ... aDf ] ò. In similar structures in which the pronoun does not c-command the launch- case assigned → WLM ing site, no problem arises: (ç¥) behin-ko dost-ne [ har lar.ke-kiiÔ ]ò [ us-keÔ ] ò (çÔ) Scrambling/A-movement: Condition C connectivity; (S)SCO every boy-u• sister-Zhh he-u• friend-u§ [[ DP-u• ...]... pron ... a[ DP-u• ...]f ] Ô Ô Ô d. a˜at˜.aa scolded case assigned → no WLM ‘For every boy x, x’s friend scolded x’s sister.’ • SSCO (ç ) Assuming that quantied DPs qualify as R-expression under the binding theory, Structure for (ç¥): No Condition C violation we also derive the observation, repeated here as (çò), that scrambling is subject to SSCO, like English A-movement and unlike English A-movement. DP ⋯ (çò) * behin-ko da˜at˜ aa [ har lar.ke-kiiÔ ]ò us-neÔ ò . . every boy-u• sister-Zhh he-u§ scolded har lar.ke-kiiÔ behin-ko DP ⋯ ‘For every boy x, x scolded x’s sister.’

us-keÔ dost-ne DP d. a˜at˜.aa

behin-ko har lar.ke-kiiÔ

• Summary We proposed that scrambling involves A-movement of a case-marked DP. ⪧ Because it targets an A-position, it feeds pronominal binding → no WCO

â ⪧ Because it targets case-marked DPs, its copy necessarily contains an NP • ree types of movement restrictor → Condition C connectivity + SSCO Ô. Pre-case, creates new binding options: English A-movement ¾ mixed A/A-behavior → no WCO, no SSCO, no Condition C connectivity ò. Post-case, creates new binding options: Hindi scrambling ç.ò.ç Extension to reciprocals → no WCO, but SSCO and Condition C connectivity • Our account extends to a restriction on the distribution of reciprocals. ey ç. Post-case, does not create new binding options: English A-movement may not be bound as a result of movement if the reciprocal c-commands the → WCO, SSCO, and Condition C connectivity launching site. • Our analysis does not treat scrambling as a third primitive type of movement (çâ) Restriction on reciprocal binding with an arbitrary set of properties. Rather, we have explored the possibility that

a. [ Rina aur Mina ]-koÔ [ ek duusre-kiiÔ maao-ne˜ ] Ô WCO on the one hand and SSCO and Condition C connectivity on the other Rina and Mina -Zhh each other-u• mothers-u§ are conditioned by independent properties of a movement type, which happen d. a˜at˜.aa to largely correlate in English. scolded – WCO ‘Rina and Mina , each other’s mothers scolded.’ Ô Ô WCO is conditioned by the A/A-nature of the landing site. b. * da˜at˜ aa [ Rina aur Mina ]-koÔ ek duusre-neÔ Ô . . – SSCO + Condition C Rina and Mina -Zhh each other-u§ scolded SSCO and Condition C connectivity are conditioned by case: movement gives

‘Rina and MinaÔ, each otherÔ scolded.’ rise to SSCO and Condition C connectivity if it applies to case-marked DPs. is is the case for English A-movement, and Hindi scrambling demonstrates c. [ Rina aur Mina ]-neÔ ek duusre-koÔ d. a˜at˜.aa Rina and Mina -u§ each other-Zhh scolded that it can also be the case for A-movement. ‘Rina and Mina scolded each other .’ Ô Ô (ç—) English Hindi English A-movement scrambling A-movement Scrambling′? • is restriction now follows as a Condition C ešect: e launching site contains a full copy, which incurs a Condition C ešect if it is c-commanded by the Type of landing site AA A A reciprocal. WCO NN Y Y (çÞ) *[Rina and Mina] ... each other ... a[Rina and Mina] f ... Ô Ô Ô (S)SCO NY Y N possessor Condition C N NY Y ¥ Some implications connectivity feeds case Y N N Y ¥.Ô Typology of movement types • • We investigated an apparently paradoxical constellation of properties of Hindi A fourth type of movement? As it stands, our account gives rise to the expectation that there might also be a scrambling relative to the A/A-distinction: scrambling behaves like A-movement fourth movement type ( ′ in (ç—)): A-movement that does feed case in not being subject to WCO, but like A-movement in being subject to Condi- scrambling assignment is predicted to not allow binding from the landing site, but it should tion C connectivity and (S)SCO. show Condition C obviation with respect to the launching site. Is this the case?

Þ • Implications for the A/A-nature of scrambling (¥ò) a. Anu-koÔ Sita-ne Ô dekhaa Anu-Zhh Sita-u§ saw – Webelhuth(ÔÀ—À), Dayal(ÔÀÀ¥), and others: scrambling targets a mixed position that simultaneously has A- and A-properties. ‘Anu, Sita saw.’

– Mahajan(ÔÀÀý): scrambling can be either A- or A-movement (but not simul- b. Anu-koÔ Sita-ne ⟨Anu-koÔ⟩ dekhaa taneously both), and that surface scrambling congurations are ambiguous Anu-Zhh Sita-u§ saw as to the movement type involved. µ Condition C must be blind to higher copies, even if they are in an A-position. • From one perspective, the evidence here argues for treating scrambling as a third type of movement that cannot be reduced to either English A- or A-movement, (¥ç) Revised Condition C thus providing support for Webelhuth’s and Dayal’s position. An R-expression must be globally A-free w.r.t. distinct DPs, but does not have to be A-free w.r.t. higher occurrences of itself. • However, by recognizing that WCO and (S)SCO correlate with dišerent aspects of a movement type, we obtain a more ne-grained typology of movement that • Deriving (¥ç) obviates the need to postulate a new type of movement as a theoretical primitive. We can make sense of this in a multidominance system. Assuming that scram- bling targets SpecTP, such a system attributes to (¥ò) the structure in (¥¥). ¥.ò Comparison to a trace-based account of SCO (¥¥) Multidominance structure for (¥ò) • Traditional account of SCO TP A-movement leaves behind a special trace (i.e., a variable) that is subject to Condition C and must hence be globally A-free. T′

(çÀ) *Who does she like tA? Ô Ô Ô T vP • Account here On our account translates this standard treatment into a copy-theoretic frame- Sita-ne v′ work. It retains the insight that SCO is due to Condition C, but not because A-movement leaves behind a special kind of trace, but because it leaves a full v VP copy. Anu-ko V (¥ý) *WhoÔ does sheÔ like ⟨whoÔ⟩?

• Unlike the traditional treatment, a copy-theoretic conception extends to SSCO. • Here the moving element occupies two positions at the same time. Under a standard conception of c-command (such as (¥ )), Anu-ko does not c-command (¥Ô) * Whose mother does like mother ? [ Ô ] sheÔ a[whoseÔ ]f itself. (¥ ) • What about the moved element itself? C-command If the full copy of a moved DP behaves like an R-expression, then why doesn’t α c-commands β iš (i) α ≠ β, Hindi scrambling always result in a Condition C violation? (ii) α does not dominate β, and (iii) every node that dominates α also dominates β.

— (¥â) Condition C ( ý) CP An R-expression must be globally A-free. ⋯ (¥Þ) A DP is globally A-free if it is not c-commanded by a coindexed DP.

• Consequence sheÔ ⋯ e only binding relationships that are relevant are the ones between a c-commanding pronoun and the ‘trace’ of the moved XP. show DP

• Improper movement As far as we know, the only instance where the binding relationship between a which picture that MaryÔ took moved element and its trace end up doing some work is as part of an explanation during her recent trip of why improper movement is ungrammatical (May ÔÀÞÀ, Chomsky ÔÀ—Ô). • e puzzle also arises for Hindi scrambling, which shows Condition C amnesty (¥—) *What seems (that) A Mary read A ? Ô [ tÔ tÔ ] with R-expressions inside relative clauses.

• e moved DP what ends up in an A-position, from which it A-binds the variable • To rescue the late-merge account, we adopt proposals by Johnson(òýÔý) and leŸ by the embedded A-movement step. is results in a Condition C violation Poole(òýÔÞ), according to which only the NP part of a DP is shared across occur- and a blocking of improper movement. Luckily for us, this is not the only way rences. is allows late merge of adjuncts. e following structure is simplied to block improper movement (Müller & Sternefeld ÔÀÀç, Williams òýýç, Abels from Poole(òýÔÞ). òýýÞ, Keine òýÔÀ). ( Ô) CP

¥.ç Multidominance and Late Merge ⋯ DP • A multidominance conception of movement seems to undermine (Wholesale)

Late Merge because late-merging material into the higher occurrence is identical sheÔ ⋯ D NP to late-merging it into the lower occurrence. is should create Condition C which connectivity. show DP CP

(¥À) [ Which picture that MaryÔ took during her recent trip ]ò did sheÔ show D N that Mary took ò to Alex? Ô the picture during her recent trip

¥.¥ Taking stock

• Landing site of scrambling has A-properties → no WCO

• Launching site of scrambling has A-properties due to case → SSCO + Condi- tion C connectivity

⪧ Scrambling can feed pronominal binding, but not if the pronoun c-commands the launching site

À • Scrambling shows mixed A/A-behavior that is not reducible to an Mahajan, Anoop. ÔÀÀý. e A/A-bar distinction and movement theory. Ph.D. disserta- between typical A-movement and A-movement. tion, MIT, Cambridge, MA. May, Robert. ÔÀÞÞ. e grammar of quantication. Ph.D. dissertation, MIT, Cambridge, • Our conclusion suggested a particular multidominance conception of move- MA. ment. May, Robert. ÔÀÞÀ. Must Comp-to-Comp movement be stipulated? Linguistic Inquiry Ôý: Acknowledgments ÞÔÀ–Þò . We thank audiences at FASAL — (Wichita State University), UMass Amherst, and NELS ¥À May, Robert & Alan Bale. òýýâ. Inverse linking. In: e Blackwell Companion to , (Cornell). In particular, we thank Veneeta Dayal, Ashwini Deo, Bilge Palaz, Martin Salzmann, Vol. II, ed. by Martin Everaert & Henk van Riemsdijk, Oxford: Blackwell, pp. Giorgos Spathas, and Gary oms for comments, suggestions, and data. âçÀ–ââÞ. Müller, Gereon & Wolfgang Sternefeld. ÔÀÀç. Improper movement and unambiguous binding. Linguistic Inquiry ò¥: ¥âÔ– ýÞ. Abels, Klaus. òýýÞ. Towards a restrictive theory of (remnant) movement. In: Linguis- Poole, Ethan. òýÔÞ. Movement and the semantic type of traces. Ph.D. dissertation, tic Variation Yearbook Þ, ed. by Jeroen van Craenenbroeck & Johan Rooryck, University of Massachusetts, Amherst, MA. Amsterdam: John Benjamins, pp. ç–Ôòý. Postal, Paul. ÔÀÞÔ. Cross-Over Phenomena. New York: Holt, Rinehart and Winston. Chomsky, Noam. ÔÀ—Ô. Lectures on Government and Binding. Dordrecht: Foris. Postal, Paul. ÔÀÀç. Remarks on weak crossover ešects. Linguistic Inquiry ò¥: çÀ– â. Chomsky, Noam. ÔÀÀç. A minimalist program for linguistic theory. In: e View from Reinhart, Tanya. ÔÀ—ç. and Semantic Interpretation. London: Croom Helm. Building òý, ed. by Ken Hale & Samuel Jay Keyser, Cambridge, MA: MIT Press, Ruys, Eddy. òýýý. Weak crossover as a phenomenon. Linguistic Inquiry çÔ: Ôç– pp. Ô– ò. çÀ. Dayal, Veneeta Srivastav. ÔÀÀ¥. Binding facts in Hindi and the scrambling phenomenon. Sar, Ken. ÔÀÀÀ. Vehicle change and reconstruction in A-chains. Linguistic Inquiry çý: In: eoretical Perspectives on Word Order in South Asian Languages, ed. by Miriam —Þ–âòý. Butt, Tracy Holloway King & Gillian Ramchand, Stanford: CSLI, pp. òçÞ–òâò. Sauerland, Uli. ÔÀÀ—. On the making and meaning of chains. Ph.D. dissertation, MIT, Fox, Danny. ÔÀÀÀ. Reconstruction, binding theory, and the interpretation of chains. Cambridge, MA. Linguistic Inquiry çý: Ô Þ–ÔÀâ. Sauerland, Uli. òýý¥. e interpretation of traces. Natural Language Ôò: Gurtu, Madhu. ÔÀÀò. Anaphoric Relations in Hindi and English. New Delhi: Munshiram âç–ÔòÞ. Manoharlal. Takahashi, Shoichi & Sarah Hulsey. òýýÀ. Wholesale late merger: Beyond the A/A Higginbotham, James. ÔÀ—ý. and bound variables. Linguistic Inquiry ÔÔ: âÞÀ– distinction. Linguistic Inquiry ¥ý: ç—Þ–¥òâ. Þý—. van Riemsdijk, Henk & Edwin Williams. ÔÀ—Ô. NP structure. e Linguistic Review Ô: Johnson, Kyle. òýÔý. Towards deriving dišerences in how wh movement and QR are ÔÞÔ–òԗ. pronounced, Ms., University of Massachusetts, Amherst. To appear in Lingua. Wasow, omas. ÔÀÞò. Anaphoric relations in English. Ph.D. dissertation, MIT, Cam- Keine, Stefan. òýԗ. Case vs. positions in the locality of A-movement. Glossa ç: Ôç—. bridge, MA. Keine, Stefan. òýÔÀ. Selective opacity. Linguistic Inquiry ý: Ôç–âò. Webelhuth, Gert. ÔÀ—À. Syntactic saturation phenomena and the modern Germanic Keine, Stefan & Ethan Poole. òýԗa. Interpreting long scrambling in Hindi-Urdu. In: language. Ph.D. dissertation, University of Massachusetts, Amherst. Proceedings of the ¥—th Annual Meeting of the North East Linguistic Society (NELS Williams, Edwin. òýýç. Representation eory. Cambridge, MA: MIT Press. ¥—), Vol. ò, ed. by Max Nelson & Sherry Hucklebridge, Amherst, MA: GLSA, pp. Ôý –Ôԗ. Keine, Stefan & Ethan Poole. òýԗb. Not all reconstruction ešects are syntactic, Ms., University of Southern California and University of California, Los Angeles. Lebeaux, David. ÔÀ——. Language acquisition and the form of the grammar. Ph.D. dis- sertation, University of Massachusetts, Amherst, MA. Lebeaux, David. òýýý. Language Acquisition and the Form of the Grammar. Amsterdam: John Benjamins.

Ôý Appendix A: Scrambling in Turkish b. Clitic doubling → no WCO [ I mitera tuÔ ] to-ajapai (to) kathe pedhiÔ. • e following data are due to Bilge Palaz (p.c.). the mother his him-loves the every boy ‘For every boy , ’s mother loves .’ ( ò) Scrambling feeds binding x x x

Her çocuğ-uÔ [ proÔ~ò kardeş-i ] Ô azarladı. c. Clitic doubling + fronting → no WCO every child-acc sibling-£™«« scolded (To) kathe pedhiÔ to-ajapai [ i mitera tuÔ ]. ‘For every child x, {x’s~somebody else’s} sibling scolded x.’ the every boy him-loves the mother his ‘For every boy x, x’s mother loves x.’ ( ç) Binding by possessors

[ Her çocuğ-unÔ kardeş-u ] onuÔ~ò azarladı. ( Þ) Binding by possessors every child-u• sibling-£™«« him/her scolded ?[ I mitera (tu) kathe mathitiÔ ] tonÔ-ajapai poli. ‘For ever child x, x’s sibling scolded {x~someone else}.’ the mother the every student him-loves very ‘Every student’s mother loves him a lot.’ ( ¥) Scrambling + binding by possessor

[ Her çocuğ-unÔ kardeş-i-ni ] ò [ proÔ~ò arkadaş-ı ] ò ( —) Clitic doubling (+ scrambling) feeds binding if pronoun does not c-command every child-u• sibling-£™««-Zhh friend-£™«« trace azarladı. a. WCO scolded *Vathmolojise [ o dhaskalos tuÔ ] [ to jrapto (tu) kathe ‘For every child x, x’s friend scolded x’s sibling.’ graded the teacher his the exam the every (unbound readings and binding by the j index also possible) mathitiÔ ]. student ( ) No binding if pronoun c-commands trace intended: ‘For every student x, x’s teacher graded x’s exam.’ [ Her çocuğ-unÔ kardeş-i-ni ] ò o*Ô~ò ò azarladı. every child-u• sibling-£™««-Zhh he/she scolded b. Clitic doubling → no WCO

‘For every child x, he/she scolded x’s sibling.’ ?To-vathmolojise [ o dhaskalos tuÔ ] [ to jrapto (tu) kathe NOT: ‘For every child x, x scolded x’s sibling.’ it-graded the teacher his the exam the every

mathitiÔ ]. student Appendix B: Greek ‘For every student x, x’s teacher graded x’s exam.’ • e following data are due to Giorgos Spathas (p.c.). c. Clitic doubling + fronting → no WCO [ To jrapto (tu) ] -vathmolojise [ o ( â) kathe mathitiÔ to Clitic doubling (+ fronting) feeds binding the exam the every student it-graded the a. WCO dhaskalos tuÔ ]. *[ I mitera tuÔ ] ajapai (to) kathe pedhiÔ. teacher his the mother his loves the every boy ‘For every student x, x’s teacher graded x’s exam.’ ‘His mother loves every boy.’

ÔÔ ( À) No binding if pronoun c-commands trace

*[ tin kori (tu) kathe anthropuÔ ] tin-ajapai proÔ. the daughter the every man her-loves intended: ‘For every man x, x loves x’s daughter.’

Ôò