Multiple scrambling, wh-movement, and Relativized Minimality
Multiple scrambling, wh-movement, and Relativized Minimality (2) a. *Kogo1 oni skazali [čto ne videli t1]? who.ACC they said.PL that NEG saw.PL „Who did they say (that) they didn‟t see?‟
Rebecca Shields University of Wisconsin-Madison [email protected] WIGL 9 – April 2, 2011
b. Ego1 oni skazali [čto ne videli t1]? him.ACC they said.PL that NEG saw.PL „Did they say (that) they didn‟t see him?‟
1. Introduction
(3) a. *Čego1 skazali [čto t1 ne budet]? what.GEN said.PL that NEG will.be „??What did they say that there won‟t be?‟
This paper looks at what happens to Relativized Minimality in Russian and Japanese when more than one argument scrambles. There is a "now-you- seeit, now-you-don't" character to Superiority violations in this context. Sometimes, Superiority effects obtain, but sometimes they do not. I argue that the apparently fleeting character of RM with scrambling follows automatically if we adopt Rizzi's 2005 feature-based implementation of RM, plus some additionalassumptions about the visibility of features in A versus A'-positions. I also explore a Minimality-based explanation for the fact that wh-movement in Russian is (surprisingly) more highly constrained than scrambling, and than wh-movement in English and German.1
b. Voobšče pisem1 skazali [čto t1 ne budet]. Well letters.GEN said.PL that NEG will.be „Well, they said there will be no letters.‟ Zemskaja 1973 (4) a. *Gde1 skazali [čto budut t1]? where said.PL that will.be.3PL „Where did they say that they will be?‟
2. Relative freedom of LD-scrambling
b. Na Vagan’kovom1 skazali [čto budut t1]. In Vagan‟kovo said.PL that will.be.3PL „They said that they will be in Vagan‟kovo (a cemetery).‟ Zemskaja 1973
Scrambling is significantly freer than wh-movement in Russian (Müller & Sternefeld 1993). 2.1 Finite clauses
Subjunctive clause (non-finite): (5) a. Kogda1 xočeš, [čtob sneg t1 rasstajal]? when want.2SG so.that snow would.melt „When do you want the snow to melt?‟
(1) a. *Čto1 ty dumaeš‟ [čto on sdelal t1]? what you think.2SG that he did „What do you think (that) he did?‟
b. Skoree1 xoču, [čtob sneg t1 rasstajal]! Sooner I.want so.that snow would.melt „I want the snow to melt soon!‟
b. Eto1 ja dumaju [čto on sdelal t1]. that I think.1SG that he did „I think (that) he did that.‟
Internet
2.2 That-trace environments 1
Grammaticality judgments on Russian sentences listed without a source are due to: Aleksandra Galambos, Kirill Gerasimov, Marija Goretskaja, Aleksey Malyutin, Galina Malyutina, Aleksandra Pavlova, and Anton Polesskij. All sourced examples were verified with some or all of these speakers as well. Japanese judgments are from Yasuhiro Sasahira, Keiko Sasahira, and Ichiro Noguchi unless otherwise noted.
(6) Ego sestra1 govorjat [čto t1 priexala] His sister.NOM say.3PL that arrived „They say that his sister arrived.‟ 1
Zemskaja 1973
Rebecca Shields (7) Nam naša pianistka1 xvatit togo [čto t1 portit nervy]. We.DAT our pianist.NOM it‟s.enough that damages nerves „It‟s more than enough for us that our pianist is getting on our nerves.‟ Zemskaja 1973
Multiple scrambling & Minimality (13) Ja bystro1 xoču [čtoby ona (*často) t1 zavodilas‟]. I quickly want [that 3.FEM (often) started.up] „I want it [e.g., a car] to (often) start up quickly.‟
3.2 Scrambling of one argument: no RM effect 2.3 Wh islands
(14) Mašu1 ja znaju [čto Vanja ne terpit t1]. Masha I know that Vanja NEG tolerate.3SG „I know that Vanja doesn‟t tolerate Masha.‟
(8) Vam1 ne očen‟ mne nravitsja [kak eto plat‟e t1 ]. You.DAT NEG very me.DAT pleases how that dress „I don‟t very much like how that dress (suits) you.‟ Zemskaja 1973 (9) Tuda1 ja ne znaju [kak idti t1]. To.there I NEG know how to.go „I don‟t know how to go there.‟
(15) Mašu1 ja znaju [čto Vanja obyčno ne terpit t1]. Masha I know that Vanja usually NEG tolerate.3SG „I know that Vanja doesn‟t usually tolerate Masha.‟ Zemskaja 1973
So far, it looks like RM constrains scrambling in the expected way.
(10) Tebe sobaki1 nravitsja [kogda t1 v dome]? You.DAT dog.NOM.PL pleases.3SG when in house „Do you like it when the dogs are in the house?‟ attested
3.3 Scrambling multiple arguments: Sometimes yes, sometimes no (16) a.
(11) O besplatnoj gazifikatsii1 ja ne znaju, [gde my čitali t1]. About free gas I NEG know where we read „I don‟t know where we read about free gas.‟ Internet
b.
(12) Bystro1 ja ne znaju [gde možno Quickly I NEG know.1SG where possible
TY1 menja2 vižu you.NOM me. ACC I.see „I see that you love me!‟
[čto t1 ljubiš‟ t2]! that love.2SG Zemskaja 1973
MENJA2 ty1 vižu [čto t1 ljubiš‟ t2]! me. ACC you.NOM I.see that love.2SG
(17) a. Valja1 mjaso2 ja znaju [čto t1 ne est t2] Valja.NOM meat.ACC I know that NEG eat.3SG „I know that Valja doesn‟t eat meat.‟
fotki t1 napečatat‟] photographs to.print attested „I don‟t know where photographs can be printed quickly‟
b. Mjaso2 Valja1 ja znaju [čto t1 meat.ACC Valja.NOM I know that
LD-scrambling is possible out of finite clauses, wh-islands, and in thattrace contexts. It is less constrained than wh-movement.
ne est t2] NEG eat.3SG
(18) a. Vse1 obo vsex2 ja vižu, [čto t1 vsjo znajut t2]. everyone about everyone I see that everything knows „I see that everyone knows everything about everybody.‟
3. What about RM? 3.1 Adjunct scrambling: RM Effect (Shields 2007)
b. *Obo vsex2 vse1 ja vižu, [čto t1 vsjo znajut t2]. about everyone everyone I see that everything knows 2
Rebecca Shields
Multiple scrambling & Minimality
(19) a. Vse1 vsjo2 ja znaju [čto t1 ne edjat everyone everything I know that NEG eat.3PL „I know that everyone doesn‟t eat everything.‟
t2]
b. *Vsjo2 vse1 ja znaju [čto t1 ne edjat everything everyone I know that NEG eat.3PL (20) Nikto ne obižaet no one.NOM NEG bother.3SG „No one is bothering anyone.‟
(25) a. Who brought what? b. *What did who bring? It looks like there is no Superiority effect in Russian multiple whmovement, in contrast to e.g. Bulgarian (see Rudin 1988). However, (24b) is actually only possible on a D-linked reading. As is well-known, D-linked questions are not subject to Superiority (Pesetsky 1987, 2000, Shields 2008).
t2]
nikogo. no one. ACC
(26) Which present did which person bring?
(21) ??Nikto1 vižu [čto t1 ne obižaet no one.NOM see.1SG that NEG bother.3SG „I see that no one is bothering anyone.‟
nikogo.] no one. ACC
(22) Maša1 vižu [čto t1 ne obižaet Masha.NOM see.1SG that NEG bother.3SG „I see that Masha isn‟t bothering anyone.‟
nikogo.] no one. ACC
The contrast between (24-25) may be due not to parametric variation in movement types or Superiority, but to varying morphosyntactic realization of D-linking. Some facts in support of this idea: Definiteness/Specificity is not generally overtly marked in Russian: (27) Ja videla devočku I saw girl.ACC i) „I saw a girl (non-specific).‟ ii) „I saw a girl (specific).‟ iii) „I saw the girl.‟
(23) a. ??Nikto1 nikogo2 vižu [čto t1 ne obižaet t2]! no one.NOM no one. ACC see.1SG that NEG bother.3SG „I see that no one is bothering anyone.‟ b. *Nikogo2 nikto1 vižu [čto t1 ne no one. ACC no one.NOM see.1SG that NEG „I see that no one is bothering anyone.‟
Simple wh-phrases can occur in partitives in Russian, but not in English: (28) čto iz nix možno izpol‟zovat‟? What from them possible to.use „Which of them can we use?‟
obižaet t2]! bother.3SG
Superiority effect obtains when multiple QPs or NegPs are scrambled.
(29) kogda iz etix dnej lučše vstretit‟sja? When from these days best to.meet „Which of these days is it best to meet?‟
3.4 Note on apparent lack of Superiority in Russian wh-movement
(30) *What of them can we use?
(24) a. Kto čto prinosil? Who what brought b. čto kto prinosil? what who brought „Who brought what?‟
(31) *When of these days is it best to meet?
3
Rebecca Shields 3.5 Parallel Japanese examples
Multiple scrambling & Minimality Summary No RM effect: scrambling of multiple “plain” arguments RM effect: scrambling of multiple wh-phrases, quantified expressions, or negative expressions; or scrambling of one adverb across another in situ adverb
Adjunct scrambling (32) Hayakui, boku-wa [Peter-ga (*hinpan-ni) ti kuruma-o unten-suru to] Fasti I-TOP P.-NOM frequently ti car-ACC drive that sinjite-iru. believe „I believe that Peter frequently drives fast.‟
4. Approaches that do not work A. Suppose that scrambling does not involve any kind of syntactic
Multiple argument scrambling (33) a. John-ni1 sono hon-o2 [Bill-ga [Mary-ga t 1 t2 John-DAT that book-ACC Bill-NOM Mary-NOM
dependency. Bošković & Takahashi 1998: scrambling is traceless Last
Resort lowering. Incorrectly predicts no locality effects whatsoever.
watasita to] itta]. handed that said „Bill said that Mary handed John that book.‟ b.
Sono hon-o2 That book-ACC
B. Suppose that scrambling does involve a syntactic dependency. Some version of Minimality can constrain it.
John-ni1 [Bill-ga [Mary-ga t 1 t2 John-DAT Bill-NOM Mary-NOM
Some possibilities: 1. Rizzi 1990/2001 – RM with 3 structural types: A, A‟, head
watasita to] itta]. handed that said „Bill said that Mary handed John that book.‟ (Bošković & Takahashi 1998) (34)
Incorrectly predicts any intervening A‟ position to block scrambling (8-12, 15, 16-17 should all be *)
a. Dare-ni1 nani-o2 [John-ga [Tanaka-sensee-ga t1 t2 who-DAT what-ACC J.-NOM T.-teacher-NOM
2. Chomsky 1993 – Shortest Move + Equidistance or Chomsky 1995 – Attract Closest + Equidistance
yomaseta to] itta no]? read- CAUS that said Q 'Who did John say Professor Tanaka made read what?'
If the subject & object are Equidistant, should be able to scramble either over the other freely (18b, 19b, 23b should all be OK)
b.*Nani-o2 dare-ni1 [John-ga [Tanaka-sensee-ga t1 t2 What-ACC who-DAT J.-NOM T.-teacher-NOM yomaseta to] itta no]? read-CAUS that said Q
None of these are able to correctly distinguish between the RM cases and no-RM cases. Problem: all define Minimality only in terms of a particular structural configuration. Not sensitive to feature makeup.
(Richards 2001, citing Takako Aikawa, p.c.) 4
Rebecca Shields
5. Rizzi‟s 2005 refined relativized feature-based approach Relativized Minimality (Rizzi 2001) (35) (A1,…,An) is a chain iff, for 1≤i
Rebecca Shields University of Wisconsin-Madison [email protected] WIGL 9 – April 2, 2011
b. Ego1 oni skazali [čto ne videli t1]? him.ACC they said.PL that NEG saw.PL „Did they say (that) they didn‟t see him?‟
1. Introduction
(3) a. *Čego1 skazali [čto t1 ne budet]? what.GEN said.PL that NEG will.be „??What did they say that there won‟t be?‟
This paper looks at what happens to Relativized Minimality in Russian and Japanese when more than one argument scrambles. There is a "now-you- seeit, now-you-don't" character to Superiority violations in this context. Sometimes, Superiority effects obtain, but sometimes they do not. I argue that the apparently fleeting character of RM with scrambling follows automatically if we adopt Rizzi's 2005 feature-based implementation of RM, plus some additionalassumptions about the visibility of features in A versus A'-positions. I also explore a Minimality-based explanation for the fact that wh-movement in Russian is (surprisingly) more highly constrained than scrambling, and than wh-movement in English and German.1
b. Voobšče pisem1 skazali [čto t1 ne budet]. Well letters.GEN said.PL that NEG will.be „Well, they said there will be no letters.‟ Zemskaja 1973 (4) a. *Gde1 skazali [čto budut t1]? where said.PL that will.be.3PL „Where did they say that they will be?‟
2. Relative freedom of LD-scrambling
b. Na Vagan’kovom1 skazali [čto budut t1]. In Vagan‟kovo said.PL that will.be.3PL „They said that they will be in Vagan‟kovo (a cemetery).‟ Zemskaja 1973
Scrambling is significantly freer than wh-movement in Russian (Müller & Sternefeld 1993). 2.1 Finite clauses
Subjunctive clause (non-finite): (5) a. Kogda1 xočeš, [čtob sneg t1 rasstajal]? when want.2SG so.that snow would.melt „When do you want the snow to melt?‟
(1) a. *Čto1 ty dumaeš‟ [čto on sdelal t1]? what you think.2SG that he did „What do you think (that) he did?‟
b. Skoree1 xoču, [čtob sneg t1 rasstajal]! Sooner I.want so.that snow would.melt „I want the snow to melt soon!‟
b. Eto1 ja dumaju [čto on sdelal t1]. that I think.1SG that he did „I think (that) he did that.‟
Internet
2.2 That-trace environments 1
Grammaticality judgments on Russian sentences listed without a source are due to: Aleksandra Galambos, Kirill Gerasimov, Marija Goretskaja, Aleksey Malyutin, Galina Malyutina, Aleksandra Pavlova, and Anton Polesskij. All sourced examples were verified with some or all of these speakers as well. Japanese judgments are from Yasuhiro Sasahira, Keiko Sasahira, and Ichiro Noguchi unless otherwise noted.
(6) Ego sestra1 govorjat [čto t1 priexala] His sister.NOM say.3PL that arrived „They say that his sister arrived.‟ 1
Zemskaja 1973
Rebecca Shields (7) Nam naša pianistka1 xvatit togo [čto t1 portit nervy]. We.DAT our pianist.NOM it‟s.enough that damages nerves „It‟s more than enough for us that our pianist is getting on our nerves.‟ Zemskaja 1973
Multiple scrambling & Minimality (13) Ja bystro1 xoču [čtoby ona (*často) t1 zavodilas‟]. I quickly want [that 3.FEM (often) started.up] „I want it [e.g., a car] to (often) start up quickly.‟
3.2 Scrambling of one argument: no RM effect 2.3 Wh islands
(14) Mašu1 ja znaju [čto Vanja ne terpit t1]. Masha I know that Vanja NEG tolerate.3SG „I know that Vanja doesn‟t tolerate Masha.‟
(8) Vam1 ne očen‟ mne nravitsja [kak eto plat‟e t1 ]. You.DAT NEG very me.DAT pleases how that dress „I don‟t very much like how that dress (suits) you.‟ Zemskaja 1973 (9) Tuda1 ja ne znaju [kak idti t1]. To.there I NEG know how to.go „I don‟t know how to go there.‟
(15) Mašu1 ja znaju [čto Vanja obyčno ne terpit t1]. Masha I know that Vanja usually NEG tolerate.3SG „I know that Vanja doesn‟t usually tolerate Masha.‟ Zemskaja 1973
So far, it looks like RM constrains scrambling in the expected way.
(10) Tebe sobaki1 nravitsja [kogda t1 v dome]? You.DAT dog.NOM.PL pleases.3SG when in house „Do you like it when the dogs are in the house?‟ attested
3.3 Scrambling multiple arguments: Sometimes yes, sometimes no (16) a.
(11) O besplatnoj gazifikatsii1 ja ne znaju, [gde my čitali t1]. About free gas I NEG know where we read „I don‟t know where we read about free gas.‟ Internet
b.
(12) Bystro1 ja ne znaju [gde možno Quickly I NEG know.1SG where possible
TY1 menja2 vižu you.NOM me. ACC I.see „I see that you love me!‟
[čto t1 ljubiš‟ t2]! that love.2SG Zemskaja 1973
MENJA2 ty1 vižu [čto t1 ljubiš‟ t2]! me. ACC you.NOM I.see that love.2SG
(17) a. Valja1 mjaso2 ja znaju [čto t1 ne est t2] Valja.NOM meat.ACC I know that NEG eat.3SG „I know that Valja doesn‟t eat meat.‟
fotki t1 napečatat‟] photographs to.print attested „I don‟t know where photographs can be printed quickly‟
b. Mjaso2 Valja1 ja znaju [čto t1 meat.ACC Valja.NOM I know that
LD-scrambling is possible out of finite clauses, wh-islands, and in thattrace contexts. It is less constrained than wh-movement.
ne est t2] NEG eat.3SG
(18) a. Vse1 obo vsex2 ja vižu, [čto t1 vsjo znajut t2]. everyone about everyone I see that everything knows „I see that everyone knows everything about everybody.‟
3. What about RM? 3.1 Adjunct scrambling: RM Effect (Shields 2007)
b. *Obo vsex2 vse1 ja vižu, [čto t1 vsjo znajut t2]. about everyone everyone I see that everything knows 2
Rebecca Shields
Multiple scrambling & Minimality
(19) a. Vse1 vsjo2 ja znaju [čto t1 ne edjat everyone everything I know that NEG eat.3PL „I know that everyone doesn‟t eat everything.‟
t2]
b. *Vsjo2 vse1 ja znaju [čto t1 ne edjat everything everyone I know that NEG eat.3PL (20) Nikto ne obižaet no one.NOM NEG bother.3SG „No one is bothering anyone.‟
(25) a. Who brought what? b. *What did who bring? It looks like there is no Superiority effect in Russian multiple whmovement, in contrast to e.g. Bulgarian (see Rudin 1988). However, (24b) is actually only possible on a D-linked reading. As is well-known, D-linked questions are not subject to Superiority (Pesetsky 1987, 2000, Shields 2008).
t2]
nikogo. no one. ACC
(26) Which present did which person bring?
(21) ??Nikto1 vižu [čto t1 ne obižaet no one.NOM see.1SG that NEG bother.3SG „I see that no one is bothering anyone.‟
nikogo.] no one. ACC
(22) Maša1 vižu [čto t1 ne obižaet Masha.NOM see.1SG that NEG bother.3SG „I see that Masha isn‟t bothering anyone.‟
nikogo.] no one. ACC
The contrast between (24-25) may be due not to parametric variation in movement types or Superiority, but to varying morphosyntactic realization of D-linking. Some facts in support of this idea: Definiteness/Specificity is not generally overtly marked in Russian: (27) Ja videla devočku I saw girl.ACC i) „I saw a girl (non-specific).‟ ii) „I saw a girl (specific).‟ iii) „I saw the girl.‟
(23) a. ??Nikto1 nikogo2 vižu [čto t1 ne obižaet t2]! no one.NOM no one. ACC see.1SG that NEG bother.3SG „I see that no one is bothering anyone.‟ b. *Nikogo2 nikto1 vižu [čto t1 ne no one. ACC no one.NOM see.1SG that NEG „I see that no one is bothering anyone.‟
Simple wh-phrases can occur in partitives in Russian, but not in English: (28) čto iz nix možno izpol‟zovat‟? What from them possible to.use „Which of them can we use?‟
obižaet t2]! bother.3SG
Superiority effect obtains when multiple QPs or NegPs are scrambled.
(29) kogda iz etix dnej lučše vstretit‟sja? When from these days best to.meet „Which of these days is it best to meet?‟
3.4 Note on apparent lack of Superiority in Russian wh-movement
(30) *What of them can we use?
(24) a. Kto čto prinosil? Who what brought b. čto kto prinosil? what who brought „Who brought what?‟
(31) *When of these days is it best to meet?
3
Rebecca Shields 3.5 Parallel Japanese examples
Multiple scrambling & Minimality Summary No RM effect: scrambling of multiple “plain” arguments RM effect: scrambling of multiple wh-phrases, quantified expressions, or negative expressions; or scrambling of one adverb across another in situ adverb
Adjunct scrambling (32) Hayakui, boku-wa [Peter-ga (*hinpan-ni) ti kuruma-o unten-suru to] Fasti I-TOP P.-NOM frequently ti car-ACC drive that sinjite-iru. believe „I believe that Peter frequently drives fast.‟
4. Approaches that do not work A. Suppose that scrambling does not involve any kind of syntactic
Multiple argument scrambling (33) a. John-ni1 sono hon-o2 [Bill-ga [Mary-ga t 1 t2 John-DAT that book-ACC Bill-NOM Mary-NOM
dependency. Bošković & Takahashi 1998: scrambling is traceless Last
Resort lowering. Incorrectly predicts no locality effects whatsoever.
watasita to] itta]. handed that said „Bill said that Mary handed John that book.‟ b.
Sono hon-o2 That book-ACC
B. Suppose that scrambling does involve a syntactic dependency. Some version of Minimality can constrain it.
John-ni1 [Bill-ga [Mary-ga t 1 t2 John-DAT Bill-NOM Mary-NOM
Some possibilities: 1. Rizzi 1990/2001 – RM with 3 structural types: A, A‟, head
watasita to] itta]. handed that said „Bill said that Mary handed John that book.‟ (Bošković & Takahashi 1998) (34)
Incorrectly predicts any intervening A‟ position to block scrambling (8-12, 15, 16-17 should all be *)
a. Dare-ni1 nani-o2 [John-ga [Tanaka-sensee-ga t1 t2 who-DAT what-ACC J.-NOM T.-teacher-NOM
2. Chomsky 1993 – Shortest Move + Equidistance or Chomsky 1995 – Attract Closest + Equidistance
yomaseta to] itta no]? read- CAUS that said Q 'Who did John say Professor Tanaka made read what?'
If the subject & object are Equidistant, should be able to scramble either over the other freely (18b, 19b, 23b should all be OK)
b.*Nani-o2 dare-ni1 [John-ga [Tanaka-sensee-ga t1 t2 What-ACC who-DAT J.-NOM T.-teacher-NOM yomaseta to] itta no]? read-CAUS that said Q
None of these are able to correctly distinguish between the RM cases and no-RM cases. Problem: all define Minimality only in terms of a particular structural configuration. Not sensitive to feature makeup.
(Richards 2001, citing Takako Aikawa, p.c.) 4
Rebecca Shields
5. Rizzi‟s 2005 refined relativized feature-based approach Relativized Minimality (Rizzi 2001) (35) (A1,…,An) is a chain iff, for 1≤i
