Structure Building

1
NP-movement
IP
(50b')

I'
? Theta Criterion OK
VP
I will
? Case Filter OK
V'
NP John
NP t
NP the cat
V stroke
2
NP-movement
IP

I'
? Theta Criterion OK
VP
I will
? Case Filter
V'
NP John
(the cat, t) has 2 Cases
NP t
NP the cat
V stroke
John lacks Case
3
NP-movement
IP

I'
? Theta Criterion
VP
I will
? Case Filter
NP t
V'
(the cat, t) has 2 Cases
NP t
NP the cat
V stroke
note that the cat will stroke cannot mean what
the cat will stroke itself means the cat cannot
be simultaneously associated with both ?-roles
of stroke
4
We will figure out later how to squeeze in both
be and stroked thats irrelevant for now.
NP-movement
(86)
IP

THE PASSIVE
I'
? Theta Criterion ??
VP
I will
? Case Filter ??
NP t
V'
NP t
NP the cat
V be stroked
the cat is clearly the Patient of stroke, so must
originate as the complement of stroke, and move
to subject position
but how does the cat manage to move to subject
position without violating the Theta Criterion
and the Case Filter?
5
Passive absorbs the external ?-role Passive
absorbs accusative Case.
NP-movement
(86)
IP
no ?-role

THE PASSIVE
I'
? Theta Criterion OK
VP
I will
no Case
? Case Filter OK
NP t
V'
NP t
NP the cat
V be stroked
the cat is clearly the Patient of stroke, so must
originate as the complement of stroke, and move
to subject position
but how does the cat manage to move to subject
position without violating the Theta Criterion
and the Case Filter?
6
Recall NP-movement is driven by the search for
Case.
NP-movement
Passive does not always promote the internal
argument to subject
it does in the cat was stroked (by John), but it
does not in (88a)
(88) a. John is believed IP t to have stroked
the cat b. it is believed IP John to have
stroked the cat c. it is believed CP that
John stroked the cat

here John is the external argument of the lower
predicate, normally dependent for Case on the
matrix verb (ECM)
ECM fails when the matrix verb is passivised
Passive absorbs the verbs ability to assign
structural accusative Case
7
Because seem must trigger raising to subject,
it is called a raising verb
NP-movement
the triple in (89) looks exactly like the one in
(88)
except that the matrix verb in (89) is not a
passive participle
the verb seem is born with the same handicaps
that the passive participle of believe got via an
operation (Passive)
(88) a. John is believed IP t to have stroked
the cat b. it is believed IP John to have
stroked the cat c. it is believed CP that
John stroked the cat

(89) a. John seems IP t to have stroked the
cat b. it seems IP John to have stroked the
cat c. it seems CP that John stroked the cat

seem is inherently (by way of its lexical
specification) incapable of assigning an external
?-role and accusative Case
8
This captures the signature properties of the
Passive (cf. (88)) as well.
NP-movement
the correlation between these two handicaps is
captured by (90)
(90) Burzios Generalisation a verb that does
not assign an external ?-role cannot assign
structural accusative Case
(88) a. John is believed IP t to have stroked
the cat b. it is believed IP John to have
stroked the cat c. it is believed CP that
John stroked the cat

(89) a. John seems IP t to have stroked the
cat b. it seems IP John to have stroked the
cat c. it seems CP that John stroked the cat

seem is inherently (by way of its lexical
specification) incapable of assigning an external
?-role and accusative Case
9
NP-movement
(90) Burzios Generalisation a verb that does
not assign an external ?-role cannot assign
structural accusative Case
? seem is lexically specified as in (91)
(91) seem lt?intgt, ___ IP/CP
(89) a. John seems IP t to have stroked the
cat b. it seems IP John to have stroked the
cat c. it seems CP that John stroked the cat

seem is inherently (by way of its lexical
specification) incapable of assigning an external
?-role and accusative Case
10
Historically, the terms unaccusative and
ergative have different denotations, but we
will use them interchangeably.
NP-movement
(90) Burzios Generalisation a verb that does
not assign an external ?-role cannot assign
structural accusative Case
? seem is lexically specified as in (91)
(91) seem lt?intgt, ___ IP/CP
? seem is not alone in being unable to assign a
?ext
? there are other verbs just like seem in only
assigning ?int
but differing from seem in being subcategorised
for NP
(92) unaccusative/ergative verbs arrive,
fall lt?intgt, ___ NP
11
NP-movement
unaccusative verbs are, by (90)(92), unable to
assign ACC
(90) Burzios Generalisation a verb that does
not assign an external ?-role cannot assign
structural accusative Case
they differ in their syntactic distribution from
the other type of single-argument (or monadic)
verbs, the unergative ones
(92') unergative verbs laugh, sneeze lt?extgt,
___
? while arrive, fall are unable to assign a ?ext
laugh, sneeze assign only a ?ext
(92) unaccusative/ergative verbs arrive,
fall lt?intgt, ___ NP
12
NP-movement
? the distinction between (92) and (92') is
syntactically real
? there are a number of ways in which
unaccusative and unergative verbs have
different syntactic distributions
? in English there are not very many airtight
reflexes of this dichotomy but there are
four tests that are relatively reliable
(92') unergative verbs laugh, sneeze lt?extgt,
___
(92) unaccusative/ergative verbs arrive,
fall lt?intgt, ___ NP
13
NP-movement
(a) prenominal attributive use of past participle
? possible with some unaccusatives, never with
unergatives

the sneezed boy
(92') unergative verbs laugh, sneeze lt?extgt,
___
the fallen leaves
(92) unaccusative/ergative verbs arrive,
fall lt?intgt, ___ NP
14
NP-movement
(b) cognate objects
? possible with most unergatives, never with
unaccusatives (with the exception of die, which
is arguably unaccusative but does allow a
cognate object)
he sneezed a tremendous sneeze
(92') unergative verbs laugh, sneeze lt?extgt,
___
he fell a tremendous fall

(92) unaccusative/ergative verbs arrive,
fall lt?intgt, ___ NP
15
NP-movement
(c) fake reflexives
? possible with most unergatives, never with
unaccusatives
he sneezed himself silly/into a coma
(92') unergative verbs laugh, sneeze lt?extgt,
___
he fell himself silly/into a coma

(92) unaccusative/ergative verbs arrive,
fall lt?intgt, ___ NP
16
Here I deliberately refer to ergatives rather
than unaccusatives for it is here that the
distinction between the two is actually manifest
arrive has no causative counterpart (she arrived
him).
NP-movement
(d) inchoative/causative alternation
? possible with ergatives, never with
unergatives (the unergative to burp in English
is the exception that proves the rule)
he sneezed she sneezed him
(92') unergative verbs laugh, sneeze lt?extgt,
___
the vase broke she broke the vase
the tree fell she felled the tree
(92) unaccusative/ergative verbs arrive,
fall lt?intgt, ___ NP
17
NP-movement
in languages like Dutch and Italian, the
difference between unaccusatives and unergatives
is manifest, moreover, in the choice of auxiliary
in the periphrastic perfect
? unergatives never select be, unaccusatives
usually do
hij heeft/is gelachen he is/has laughed
(92') unergative verbs laugh, sneeze lt?extgt,
___
hij is/heeft gevallen he is/has fallen
(92) unaccusative/ergative verbs arrive,
fall lt?intgt, ___ NP
18
Some unfinished business how to squeeze
in both be and stroked?
NP-movement
(86)
IP

I'
VP
I will
NP t
V'
NP t
NP the cat
V be stroked
be is very much like seem in fact,
traditionally seem is often called a
pseudo-copula
seem is a raising verb ? so lets analyse be as a
raising verb as well, so that we get (86')
19
The standard answer Binding Theory
NP-movement
(86')
IP

I'
VP
I will
? NP-movement here proceeds in three steps
NP t
V'
all
VP
V be
? at each step, a trace is left behind
V'
NP t
all
? this is supported by Q-Float
NP t
NP the cat
V stroked
the cats will ltallgt be ltallgt stroked
seem is a raising verb ? so lets analyse be as a
raising verb as well, so that we get (86')
Q what forces this so-called successive-cyclic
movement?
20
Locality of NP-movement
(94) a. the cat seems to have been
stroked b. the cat seems that it/there has
been stroked

? (94b) is a case of so-called superraising,
which fails why?
traces left behind by NP-movement are
anaphors, hence subject to Principle A of the
Binding Theory
in (94b), one NP-trace fails to be bound in its
local domain, hence violates BTA
21
Locality of NP-movement
IP

I'
NP it/there
VP
I -s
V'
NP t
CONTRA BT-A
VP
V have
V'
NP t
VP
V been
V'
NP t
CONCLUSION NP-movement must obey BT-A NP-traces
are (null) anaphors
NP the cat
V stroked
NP t
22
Locality of NP-movement
IP
chain (the cat, t, t, t, t)

I'
? well-formed by Theta Criterion all
landing-sites are non-? positions
VP
I to
V'
NP t
? well-formed by Case Filter only the final
landing- site is assigned Case
VP
V have
V'
NP t
VP
V been
? lets run through the derivation of the
grammatical case in (94a)
V'
NP t
NP the cat
V stroked
NP t
(94a) the cat seems to have been stroked
23
NP-movement Summary
moved element NP (hence the name,
NP-movement)
movement is obligatory the trigger is Case
the movement stretch is constrained by
BTA NP-trace anaphor
the extraction-site is an NP-position receiving
a ?-role
the landing-site is an NP-position lacking a
?-role
the landing-site is not necessarily a Case
position (intermediate landing-sites never are)
but the final landing-site of NP-movement of
overt NPs must be a Case position NP-movement
stops once Case is found
24
PRO and Control Theory
IP

IP needs a specifier filled by what?
I'
??
VP
I to
V'
??
PP at midnight
V'
stroke has a ?ext to assign but what to?
V stroke
NP the cat
? subject of matrix clause
to stroke the cat at midnight
? subject of infinitival clause
??
(97) to stroke the cat at midnight is not a
good idea
25
PRO and Control Theory
(98a)
governing category for himself
IP
IP has a filled specifier

I'
VP
I to
? ?-Criterion OK
V'
NP John
NP t
? BTA OK
NP himself
V kill
? EPP OK
(98) a. for John to kill himself/oneself would
be a shame b. to kill oneself/himself would
be a shame
26
NB since the null subject has no antecedent, it
isnt an NP-trace
PRO and Control Theory
(98b)
governing category for oneself
IP
IP has a filled specifier

I'
VP
I to
? ?-Criterion OK
V'
NP PRO
NP t
? BTA OK
NP oneself
V kill
? EPP OK
PRO here has arbitrary reference
PRO satisfies the EPP because it occupies
SpecIP the fact that PRO is inaudible does not
subtract from this
(98) a. for John to kill himself/oneself would
be a shame b. to kill oneself/himself would
be a shame
27
PRO and Control Theory
? PRO in (98b) behaves as if it were a pronoun
? PRO in (99) behaves as if it were an anaphor
NB the subject of the infinitival clause in (99)
must be PRO
it cannot be an NP-trace, unlike in (99'), a
raising construction
seem does not assign an external ?-role, so
raising is okay
try does assign an external ?-role, so raising is
not okay
? raising in (99) would violate the Theta
Criterion!
(99') John seems t to have killed
himself/oneself
(99) John tried PRO to kill himself/oneself
(98) a. for John to kill himself/oneself would
be a shame b. PRO to kill oneself/himself
would be a shame
28
PRO and Control Theory
? PRO in (98b) behaves as if it were a pronoun
? PRO in (99) behaves as if it were an anaphor
so PRO seems to have a schizophrenic character
it seems to be simultaneously anaphoric and
pronominal
(100) PRO is a pronominal anaphor anaphoric,
pronominal
? it follows that PRO must not have a governing
category
(101) PRO Theorem PRO is not governed
(99) John tried PRO to kill himself/oneself
(98) a. for John to kill himself/oneself would
be a shame b. PRO to kill oneself/himself
would be a shame
29
PRO and Control Theory
the PRO Theorem holds at S-structure
? if the PRO Theorem held at D-structure, (102)
would be
since the PRO Theorem holds at S-structure
and since the PRO Theorem is a theorem of Binding
Theory
? it follows that the Binding Theory holds at
S-structure
(101) PRO Theorem PRO is not governed
(102) PRO to be killed t by a sniper is not
nice
30
PRO and Control Theory
since PRO must not be governed (by (101))
? the structure of (99) cannot involve
IP-complementation
we need something to shield PRO from being
governed
? a structural layer between V and PRO would do
the job
on the assumption that V cannot govern through 2
XPs
(101) PRO Theorem PRO is not governed

(99) John tried IP PRO to be killed t by a
sniper
John tried CP C IP PRO to be killed t by a
sniper
31
PRO and Control Theory
since PRO must not be governed (by (101))
? it cannot receive structural accusative Case
PRO is not necessarily Caseless, however
? it may receive structural Case under Spec-Head
agreement
(see work on Icelandic, and also the null Case
idea)
(101) PRO Theorem PRO is not governed

(99) John tried IP PRO to be killed t by a
sniper
John tried CP C IP PRO to be killed t by a
sniper
32
PRO and Control Theory
Q what is the nature of the null subject in
(103b)?
? could it be PRO?
if the subject position of a finite root clause
is ungoverned
? there is nothing particularly wrong with ec in
(103b) PRO
the subject position of a finite non-root clause
is governed
BUT
and the subject of a finite non-root clause can
be null (103c)
? so at least ec2 in (103c) / PRO rather, ec2
in (103c) pro
this analysis generalises to all null subjects of
finite clauses
(101) PRO Theorem PRO is not governed
(103) a. Gianni ha parlato
b. ec ha parlato
c. ec1 credo che ec2 ha parlato
33
NP Typology
Up next!
variable

34
A- vs A'-movement
(94) a. the cat seems to have been
stroked b. the cat seems that it/there has
been stroked

NP-movement is also called A-movement its
landing-site is an A-position
A-positions are members of the set complement of
lex. cat., specifier of lex. cat., SpecIP
35
A- vs A'-movement
not all phrasal movement is A-movement not all
phrasal landing-sites are A-positions

the specifier position of CP is not a member of
the set complement of lex. cat., specifier of
lex. cat., SpecIP, hence is not an A-position
non-A-positions (phrasal positions that are
not A-positions) are called A'-positions
A'-positions SpecCP, adjunction positions
36
A- vs A'-movement

the specifier position of CP is not a member of
the set complement of lex. cat., specifier of
lex. cat., SpecIP, hence is not an A-position
(105) a. I would prefer CP for John to be
honest b. I think CP that John should be
honest
(106) a. ik vraag me af CP wie dat Jan gezien
heeft I wonder who that John seen has b. I
wonder CP who (that) John has seen
(107) a. IP John stroked a cat b. CP which
cat did IP John stroke?
37
A- vs A'-movement

the wh-phrase which cat does not originate
in SpecCP it is the internal argument of stroke
which cat moves at S-structure into SpecCP ? an
instance of A'-movement
(107) a. IP John stroked a cat b. CP which
cat did IP John stroke?
38
A- vs A'-movement

CP
C'
IP

C
I'
VP
I did
I tI
V'
NP tNP
NP John
NOT AN NP-TRACE/ANAPHOR
NP wh cat
V stroke
NP tWH
not bound in its governing category
39
A- vs A'-movement
could the trace of A'-movement (A'-bound trace
or variable) be a pronoun?

? to see this, we have to look beyond
mono- clausal cases of A'-movement
(112) a. Johni said hei/j saw Billk b. whoi
did Johni say ti saw Billk? c. whoj did Johni
say tj saw Billk?

Strong Crossover
(113) a. hei said Billk saw himi b. whoi did
hei say Billk saw ti? c. whoi ti said Billk
saw himi?

40
A- vs A'-movement
(112b) and (113b) would wrongly be ruled in if an
A'-bound trace/variable were a pronoun

the trace of A'-movement is NOT a pronoun
(112) a. Johni said hei/j saw Billk b. whoi
did Johni say ti saw Billk? c. whoj did Johni
say tj saw Billk?

Strong Crossover
(113) a. hei said Billk saw himi b. whoi did
hei say Billk saw ti? c. whoi ti said Billk
saw himi?

41
A- vs A'-movement
(112b) and (113b) will correctly be ruled out if
an A'-bound trace/variable is an R-expression

the trace of A'-movement is an R-expression
SCO follows from Principle C of Binding Theory
(112) a. Johni said hei/j saw Billk b. whoi
did Johni say ti saw Billk? c. whoj did Johni
say tj saw Billk?

Strong Crossover
(113) a. hei said Billk saw himi b. whoi did
hei say Billk saw ti? c. whoi ti said Billk
saw himi?

42
A- vs A'-movement

the trace of A'-movement is an R-expression
variables MUST NOT be A-bound (BT-C)
on the other hand
variables MUST be A'-bound (by definition, variabl
es are operator-bound elements)
(115) a. which cati did Johnk stroke ti? b. did
Johnk stroke ti?

43
A- vs A'-movement

since
the trace of A'-movement is an R-expression
any locality condition on A'-movement CANNOT be
derived from Binding Theory
unlike in the case of A-movement (where the ban
on superraising derives from BT-A)
(94) a. the cat t seems t to t have t been t
stroked t b. the cat t seems that it/there t has
t been t stroked t

44
A- vs A'-movement

since
the trace of A'-movement is an R-expression
any locality condition on A'-movement CANNOT be
derived from Binding Theory
Q are there any locality conditions on A'-movt?
(108) a. who did he know that he had seen
t? b. who did he know when he had seen t?

45
A- vs A'-movement

since
the trace of A'-movement is an R-expression
any locality condition on A'-movement CANNOT be
derived from Binding Theory
Q are there any locality conditions on A'-movt?
(109) a. who did he claim that he had seen
t? b. who did he hear the claim that hed seen t?

46
A'-movement Subjacency
(108) a. who did he know that he had seen
t? b. who did he know when he had seen t?


? the Wh-Island Constraint
(109) a. who did he claim that he had seen
t? b. who did he hear the claim that hed seen t?

? the Complex NP Constraint
these two constraints can be subsumed under (110)
(110) Subjacency Condition movement must not
cross more than one bounding node at a time
bounding nodes (English) IP, NP
47
A'-movement Subjacency
(108) a. who did he know that he had seen
t? b. who did IP he know when IP he had seen
t?


? the Wh-Island Constraint follows directly
(109) a. who did he claim that he had seen
t? b. who did IP he hear NP the claim that IP
t?

? the Complex NP Constraint follows directly
but how do the a-sentences obey Subjacency (110)?
(110) Subjacency Condition movement must not
cross more than one bounding node at a time
bounding nodes (English) IP, NP
48
A'-movement Subjacency
(108) a. who did IP he know that IP he had seen
t? b. who did IP he know when IP he had seen
t?


? the a-sentences are problematic as depicted here
(109) a. who did IP he claim that IP he had
seen t? b. who did IP he hear NP the claim
that IP t?

? so we need to revise their derivations
but how do the a-sentences obey Subjacency (110)?
(110) Subjacency Condition movement must not
cross more than one bounding node at a time
bounding nodes (English) IP, NP
49
A- vs A'-movement
IP
recall

I'
VP
I to
V'
NP t
VP
V have
SUCCESSIVE-CYCLIC MOVEMENT
V'
NP t
VP
V been
V'
NP t
(94a) the cat seems to have been stroked
NP the cat
V stroked
NP t
50
CP
IP
A- vs A'-movement
CP
NP ti
C'
IP
C that
I'
VP
I -ed
SUCCESSIVE-CYCLIC MOVEMENT
V'
NP tj
VP
V have
V'
NP tj
NP hej
(109a) who did he claim that he had seen?
NP whoi
V seen
NP ti
51
CP
IP
A- vs A'-movement
CP
replay
NP ti
C'
IP
C that
I'
NP hej
VP
I -ed
SUCCESSIVE-CYCLIC MOVEMENT
V'
NP tj
VP
V had
V'
NP tj
(109a) who did he claim that he had seen?
NP whoi
V seen
NP ti
52
CP
IP
A- vs A'-movement
CP
whether
C'
IP
C
I'
NP hej
VP
I -ed
V'
NP tj
VP
V had
V'
NP tj
who did he wonder whether he had seen?
NP whoi
V seen
NP ti
53
A- vs A'-movement
the trace of A-movement (NP-trace) must be bound
within its governing category (BT-A)

the trace of A'-movement (variable) must
be locally A'-bound (Subjacency)
BUT it is not sufficient to only require of
A'-bound traces that they be locally A'-bound
see Comp-trace effects
(116) a. who do you think t (that) John kissed
t? b. who do you think t (that) t kissed Mary?
54
A- vs A'-movement
the trace of A-movement (NP-trace) must be bound
within its governing category (BT-A)

the trace of A'-movement (variable) must
be locally A'-bound (Subjacency)
BUT it is not sufficient to only require of
A'-bound traces that they be locally A'-bound
see Comp-trace effects
(117) a. who would you prefer t (for) John to
kiss t? b. who would you prefer t (for) t to
kiss Mary?
55
Proper Government
The Empty Category Principle (ECP) a trace must
be properly governed
(i) ?-government
government by a head that ?-marks you
(ii) antecedent-government
having a local c-commanding antecedent
where local means in conformity with
minimality
56
Proper Government
whoi do you think ti (that) John kissed ti?
?-govt
57
Proper Government
whoi do you think ti (that) John kissed
ti? whoj do you think tj that tj kissed Mary?
?-govt
not ?-governed, and not antecedent-governed
either ECP
58
Proper Government
whoi do you think ti (that) John kissed
ti? whoj do you think tj that tj kissed
Mary? whoj do you think tj C0 tj kissed Mary?
?-govt
antecedent-government succeeds
null C is not a governor (by hypothesis)
59
Proper Government
the idea that null C is not a governor we
came across before, in the discussion of control
in- finitives in complement position null C
shields PRO from government by the matrix verb
whoj do you think tj C0 tj kissed Mary?
John tried CP C0 IP PRO to win the race
antecedent-government succeeds
null C is not a governor (by hypothesis)
60
Proper Government
but for finite null C the idea that it is not a
gov- ernor isnt innocuous we dont want the
SpecIP position of a complementiserless
embedded finite clause to be eligible for
insertion of PRO (recall that PRO can in
principle receive NOM Case)
whoj do you think tj C0 tj kissed Mary?

Johni said CP C0 IP PROi would win the race
antecedent-government succeeds
null C is not a governor (by hypothesis)
61
Proper Government
??
whoi do you wonder whenk John kissed ti tk?
?-govt
IP
IP
this sentence satisfies the ECP, but it violates
Subjacency
Subjacency violations are less severe than ECP
violations
62
Proper Government
whoi do you wonder whenk John kissed ti tk?
adjunct positions are not governed by any head
recall in this connection the grammaticality
of PRO as the subject of an adjunct small clause
IP Johni I' I VP ti V' V' left the room AP
PROi A' drunk
63
Proper Government
whenk do you think tk' that John kissed her
tk?
adjunct positions are not governed by any head
the complementiser that does not interfere
with proper government of the original trace
there is no that-t effect for adjunct extraction
64
Proper Government
whenk do you think tk' that John kissed her
tk?
?
adjunct positions are not governed by any head
it is not at all obvious how the intermediate tk'
is properly antecedent-governed ? see Syntax II
in Syntax I, well ignore the ECP for
intermediate ts
65
Proper Government
whenk do you wonder whoi John kissed ti tk?
ungrammatical on a downstairs reading of when
i.e., a reading on which the sentence enquires
about the time at which the kissing took place
66
Proper Government
whenk do you wonder whoi John kissed ti tk?
ungrammatical on a downstairs reading of when
i.e., a reading on which the sentence enquires
about the time at which the kissing took place
67
Proper Government
whenk do you wonder whoi John kissed ti tk?
BUT grammatical on an upstairs reading of when
i.e., a reading on which the sentence enquires
about the time at which the wondering took place
68
Proper Government
whenk do you wonder whoi John kissed ti tk?
BUT grammatical on an upstairs reading of when
i.e., a reading on which the sentence enquires
about the time at which the wondering took place
69
Proper Government
whoj do you wonder whenk tj kissed Mary tk?
IP
IP
this sentence violates the ECP for the subject
trace in SpecIP
and of course it also violates Subjacency two
IPs are crossed in one fell swoop
70
Proper Government
??
whoi do you wonder whenk John kissed ti tk?
?-govt
IP
IP
by comparison, this sentence is less bad it
violates Subjacency but not the ECP
71
Conditions on Movement
the documents CP which IP I filed ec
72
Conditions on Movement
the documents CP which IP I filed ec the
documents CP that IP I filed ec
73
Conditions on Movement
the documents CP which IP I filed ec the
documents CP that IP I filed ec the
documents CP IP I filed ec
74
Conditions on Movement
??
the documents CP which IP I filed ec
IP I dont remember CP whether
75
Conditions on Movement
??
the documents CP which IP I filed ec
IP I dont remember CP whether Subjacency
effect indicates that ec trace
76
Conditions on Movement
??
the documents CP that/C0 IP I filed ec
IP I dont remember CP whether
77
Conditions on Movement
??
the documents CP that/C0 IP I filed ec
IP I dont remember CP whether Subjacency
effect indicates that ec trace
78
Conditions on Movement
??
the documents CP that/C0 IP I filed ec
IP I dont remember CP whether Subjacency
effect indicates that ec trace BUT a trace of
WHAT???
79
Conditions on Movement
the documents CP thati/Ci IP I filed ti

80
Conditions on Movement
the documents CP Opi C' that/C0 IP I filed ti

81
Empty Operator Movement
the documents CP Opi C' that/C0 IP I filed ti

whats Op, doc???
Op cannot be a new null NP-type
we had already exhausted the matrix of NP-types
for empty noun phrases!
82
Empty Operator Movement
the documents CP Opi C' that/C0 IP I filed ti

whats Op, doc???
arguably, Op PRO
PRO in the SpecCP position of a relative clause
is ungoverned
83
Empty Operator Movement
the documents CP Opi C' that/C0 IP I filed ti
? finite relatives w/ that or null
complementiser
84
Empty Operator Movement
the documents CP Opi C' C0 IP PRO to file ti
the documents CP Opi C' for IP you to
file ti ? infinitival relatives
85
Empty Operator Movement
Johni is easy CP Opi C' C0 IP PRO to please ti
Johni is easy CP Opi C' for IP him to
please ti ? tough-movt / easy-to-please
constructions
86
Empty Operator Movement
Johni is easy CP Opi C' C0 IP PROi to please
ti Johni is easy CP Opi C' for IP himi
to please ti ? tough-movt /
easy-to-please constructions
Strong Crossover!
87
Empty Operator Movement
Johni is easy IP ti to ti please ti ?
NP-movt derivation yields wrong semantics ?
moreover, NP-movt is unmotivated Case! ? and
this NP-movt is illegitimate ?-Criterion!
? tough-movt / easy-to-please constructions
88
Empty Operator Movement
Johni is easy IP ti to ti please ti ?
NP-movt derivation yields wrong semantics ?
moreover, NP-movt is unmotivated Case! ? and
this NP-movt is illegitimate ?-Criterion!
Johni is easy CP ti C' C0 IP PRO to please ti
? improper movement ? tough-movt /
easy-to-please constructions
89
Empty Operator Movement
Johni is easy CP Opi C' C0 IP PRO to please ti
? no improper movement ? ti ? A-bound
within domain of its maximal chain Johni is easy
CP ti C' C0 IP PRO to please ti ?
improper movement ? tough-movt /
easy-to-please constructions
90
Empty Operator Movement
CP which documentsi did youk file ti (i)
before CP Opi IP PROk perusing ti ? (ii)
before CP Opi IP youk could peruse ti ? ?
parasitic gap constructions
91
Empty Operator Movement
?
CP which documentsi did youk file ti
before CP ti IP PROk perusing ti ? ?
parasitic gap constructions
?
? the single-chain derivation instantiates
improper movement and violates the
?-Criterion
92
Empty Operator Movement
CP which documentsi did youk file themi
before CP ti IP PROk perusing ti ? ? the
gap in the without-adjunct parasitises on a gap
in the matrix clause ? whence the name ?
parasitic gap constructions
93
Empty Operator Movement
CP which documentsi did youk file ti before
CP IP PROk perusing themi ? ? no
parasitism in the other direction, as
expected ? parasitic gap constructions
94
Empty Operator Movement
IP the documentsi were filed ti before CP
Opi IP I could peruse ti ? parasitic gaps
are not licensed by A-movt ? this gives us a
diagnostic to tell apart A- and
A'-movement ? parasitic gap constructions
95
Empty Operator Movement
CP which documentsi ti appeared in the press
before CP Opi IP you could peruse ti ? ?
the parasitic gap cannot be c-commanded by the
real gap (the anti-c-command condition) ?
this gives us a diagnostic to locate an
A'-trace ? parasitic gap constructions
96
A- vs A'-movement
A-movement starts out in a ?-position that
lacks Case terminates in a ?'-position that is
typically but not necessarily (cf. PRO) a
Case-position may pass through Caseless
?'-positions CH (Xi, Yi, Zi)
? C
? C
? C
97
A- vs A'-movement
A'-movement of an object or ECM-subject of
SC starts out in a ?-position that has
Case terminates in a ?'-position that is
Caseless may pass through Caseless
?'-positions CH (Xi, Yi, Zi)
? C
? C
? C
98
A- vs A'-movement
A'-movement of a subject in SpecIP starts out
in a ?'-position that has Case terminates in a
?'-position that is Caseless may pass through
Caseless ?'-positions CH (Xi, Yi, Zi)
? C
? C
? C
99
A- vs A'-movement
A'-movement of an adjunct starts out in a
?'-position that lacks Case terminates in a
?'-position that is Caseless may pass through
Caseless ?'-positions CH (Xi, Yi, Zi)
? C
? C
? C
100
A- vs A'-movement
all movement terminates in ?'-positions argume
nts that want Case undergo A-movt to get it ?
A-movement starts out in a Case- less
?-position, and typically terminates in a Case
position (except in the case of
PRO) constituents undergoing A'-movement
have already satisfied their Case-requirements (
if they are subject to any) ? A'-movement termina
tes in a Caseless position
101
A- vs A'-movement
BIRDS EYE VIEW A-movement A'-movement CH
(Xi, Yi, Zi) CH (Xj, Yj, Zj)
A
A
A
A'
A'
? C
? C
? C
? C
? C
? C
102
A- vs A'-movement
the properties of A- and A'-movement do not need
to be stipulated they follow from independent
principles of the theory (?-Criterion, Case
Filter) and the needs of the moved constituents
? no different movement rules are needed
? there are no movement rules
Move a move anything anywhere
103
Overt vs Covert Movement
at this point, we have satisfied all the major
principles of the theory (?, Case, BT)
CP
C'
IP
C
but we dont have who in its scope position
I'
VP
I -s
overt-syntactic A'-movement delivers the
operator-variable structure wanted in semantics
V'
NP tj
VP
V have
V'
NP tj
NP hej
(I wonder) who he has seen
NP whoi
V seen
NP ti
which x, x a human, he has he seen x
104
Overt vs Covert Movement
semanticists represent this sentence very much
like a wh-question
QR
IP
every QP adjoins to IP
IP
but there is no wh-movement going on here
NP hej
I'
D-structure S-structure PF LF
VP
I -s
we can do this overtly
V'
NP tj
but if we dont, we must do it covertly
VP
V have
everyone, he has seen
V'
NP tj
he has seen everyone
QP everyonei
V seen
QP ti
every x, x a human, he has he seen x
105
Overt vs Covert Movement
an aside
Q why cant we do wh-movement covertly?
why isnt (I wonder) he has seen who?
grammatical?
A wh-questions need to be typed as such (clause
typing)
for English, typing a clause as a
wh-question demands overt-syntactic
wh-movement into the left periphery
(so-called echo questions such as he has seen
WHO? are a special rhetorical device, only
available in root contexts)
106
Overt vs Covert Movement
delivering the correct semantic representation
for sentences with quantifiers in them is
certainly a welcome effect of QR
and it brings the additional benefit of shedding
light on the fact that a single S-structure
string sometimes supports more than one
interpretation quantifier scope ambiguity
(131) everybody loves somebody
a. for everybody it is the case that he loves
somebody
b. there is somebody such that everybody loves
him
(132) a. IP everybodyi IP somebodyj IP ti
loves tj
b. IP somebodyj IP everybodyi IP ti loves
tj
107
Overt vs Covert Movement
delivering the correct semantic representation
for sentences with quantifiers in them is
certainly a welcome effect of QR
and it brings the additional benefit of shedding
light on the fact that a single S-structure
string sometimes supports more than one
interpretation quantifier scope ambiguity
moreover, a QR-based approach to Q-scope
ambiguity can account for the fact that (133a) is
not scopally ambiguous
? QR must target the local IP
(133) a. someone thinks that everyone loves him
b. IP somebodyi IP ti thinks that
IP everybodyj IP tj loves himi
108
Overt vs Covert Movement
delivering the correct semantic representation
for sentences with quantifiers in them is
certainly a welcome effect of QR
but we would not postulate QR in syntax unless
there were syntactic evidence for its existence
we will see that a variety of arguments for QR in
LF-syntax have been presented
(i) Antecedent-Contained Deletion (ACD)
(ii) Weak Crossover (WCO)
(iii) lack of Q-scope ambiguity in wh-movement
cases
(iv) Superiority effects in multiple wh-questions
109
Overt vs Covert Movement
(134) a. IP John VP suspected everyone that
Bill did VP ec
(134a) as it stands presents an infinite
regress problem when we try to recover the
antecedent for the elliptical VP (ec)
IP John -ed VP suspect everyone that Bill did
VP ec
IP John -ed VP suspect everyone that Bill VP
did VP suspect everyone that Bill did VP
ec
(i) Antecedent-Contained Deletion (ACD)
(ii) Weak Crossover (WCO)
(iii) lack of Q-scope ambiguity in wh-movement
cases
(iv) Superiority effects in multiple wh-questions
110
Overt vs Covert Movement
(134) a. IP John VP suspected everyone that
Bill did VP ec
b. IP everyone that Bill did VP eci IP
John VP suspected ti
we can avoid this infinite regress by QRing
the object-QP out of the suspect-VP at LF, prior
to antecedent recovery for ec
only QP-objects license ACD
IP everyone that Bill did VP eci IP John
-ed VP suspect ti
IP everyone that Bill did VP suspect tii IP
John -ed VP suspect ti
fix the index!
(i) Antecedent-Contained Deletion (ACD)
(ii) Weak Crossover (WCO)
(iii) lack of Q-scope ambiguity in wh-movement
cases
(iv) Superiority effects in multiple wh-questions
111
Overt vs Covert Movement

(138) a. whoi does hisi mother love ti?

b. hisi mother loves everyonei
b. everyonei hisi mother loves ti
c. hisi mother loves Johni
WCO a variable cannot be coindexed with a
pronoun to its left
this applies straightforwardly to (138b) as well
if QPs must undergo QR at LF, leaving a variable
behind
(i) Antecedent-Contained Deletion (ACD)
(ii) Weak Crossover (WCO)
(iii) lack of Q-scope ambiguity in wh-movement
cases
(iv) Superiority effects in multiple wh-questions
112
Overt vs Covert Movement
(135) a. four boys love two girls
as for Tom, he loves Mary and Sue
a'.
as for Dick, he loves Mary and Sue
as for Harry, he loves Mary and Sue
as for John, he loves Mary and Sue
(135a) is ambiguous (135a') 4gt2 (linear scope)
exactly 4 boys the number of girls varies from 2
to 8
(i) Antecedent-Contained Deletion (ACD)
(ii) Weak Crossover (WCO)
(iii) lack of Q-scope ambiguity in wh-movement
cases
(iv) Superiority effects in multiple wh-questions
113
Overt vs Covert Movement
(135) a. four boys love two girls
as for Tom, he loves Mary and Sue
a'.
as for Dick, he loves Sue and Jackie
as for Harry, he loves Jackie and Carol
as for John, he loves Carol and Imogen
(135a) is ambiguous (135a') 4gt2 (linear scope)
exactly 4 boys the number of girls varies from 2
to 8
(i) Antecedent-Contained Deletion (ACD)
(ii) Weak Crossover (WCO)
(iii) lack of Q-scope ambiguity in wh-movement
cases
(iv) Superiority effects in multiple wh-questions
114
Overt vs Covert Movement
(135) a. four boys love two girls
as for Tom, he loves Mary and Sue
a'.
as for Dick, he loves Christie and Jackie
as for Harry, he loves Carla and Carol
as for John, he loves Gwendolyn and Imogen
(135a) is ambiguous (135a') 4gt2 (linear scope)
exactly 4 boys the number of girls varies from 2
to 8
(i) Antecedent-Contained Deletion (ACD)
(ii) Weak Crossover (WCO)
(iii) lack of Q-scope ambiguity in wh-movement
cases
(iv) Superiority effects in multiple wh-questions
115
Overt vs Covert Movement
(135) a. four boys love two girls
as for Mary, Tom, Dick, Harry and John love her
a''.
as for Sue, Tom, Dick, Harry and John love her
(135a) is ambiguous (135a'') 2gt4 (inverse
scope)
exactly 2 girls the number of girls varies from
4 to 8
(i) Antecedent-Contained Deletion (ACD)
(ii) Weak Crossover (WCO)
(iii) lack of Q-scope ambiguity in wh-movement
cases
(iv) Superiority effects in multiple wh-questions
116
Overt vs Covert Movement
(135) a. four boys love two girls
as for Mary, Tom, Dick, Harry and John love her
a''.
as for Sue, Tom, Dick, Barry and Bubba love her
(135a) is ambiguous (135a'') 2gt4 (inverse
scope)
exactly 2 girls the number of girls varies from
4 to 8
(i) Antecedent-Contained Deletion (ACD)
(ii) Weak Crossover (WCO)
(iii) lack of Q-scope ambiguity in wh-movement
cases
(iv) Superiority effects in multiple wh-questions
117
Overt vs Covert Movement
(135) a. four boys love two girls
as for Mary, Tom, Dick, Harry and John love her
a''.
as for Sue, Bill, Bob, Barry and Bubba love her
(135a) is ambiguous (135a'') 2gt4 (inverse
scope)
exactly 2 girls the number of girls varies from
4 to 8
(i) Antecedent-Contained Deletion (ACD)
(ii) Weak Crossover (WCO)
(iii) lack of Q-scope ambiguity in wh-movement
cases
(iv) Superiority effects in multiple wh-questions
118
Overt vs Covert Movement
(135) a. four boys love two girls
b. which two girls do four boys love?
CP wh 2 girlsj C' C IP 4 boysi IP ti
love tj
while (135a) is ambiguous, (135b) is not 2gt4,
4gt2
this follows if Q-scope ambiguity is the result
of QR to IP, and wh-movement targets a position
higher than IP (SpecCP)
(i) Antecedent-Contained Deletion (ACD)
(ii) Weak Crossover (WCO)
(iii) lack of Q-scope ambiguity in wh-movement
cases
(iv) Superiority effects in multiple wh-questions
119
Overt vs Covert Movement
(135) a. four boys love two girls
b. which two girls do four boys love?
CP wh 2 girlsj C' C IP 4 boysi IP ti
love tj
c. four boys love which two girls? Chinese
just like English (135b), Chinese (135c) is
unambiguous 2gt4
this follows if wh-in-situ undergoes LF-movt to
SpecCP
(i) Antecedent-Contained Deletion (ACD)
(ii) Weak Crossover (WCO)
(iii) lack of Q-scope ambiguity in wh-movement
cases
(iv) Superiority effects in multiple wh-questions
120
Overt vs Covert Movement

(138) a. whoi does hisi mother love ti?

b'. why does hisi mother love whoi?
_at_LF whoi why hisi mother loves ti?

WCO a variable cannot be coindexed with a
pronoun to its left
this applies straightforwardly to the wh-in-situ
case as well if wh-in-situ must undergo LF-movt,
leaving a variable behind
back to WCO
(i) Antecedent-Contained Deletion (ACD)
(ii) Weak Crossover (WCO)
(iii) lack of Q-scope ambiguity in wh-movement
cases
(iv) Superiority effects in multiple wh-questions
121
Overt vs Covert Movement

(138) a. whoi does hisi mother love ti?

b'. why does hisi mother love whoi?
_at_LF whoi why hisi mother loves ti?

the idea that wh-in-situ must undergo LF-movt to
SpecCP also allows us to make sense of the
particular interpretation of multiple
wh-questions like who ate what?
pair-list readings
Mary ate pizza, Bill a steak, Sue a melon,
(i) Antecedent-Contained Deletion (ACD)
(ii) Weak Crossover (WCO)
(iii) lack of Q-scope ambiguity in wh-movement
cases
(iv) Superiority effects in multiple wh-questions
122
Overt vs Covert Movement
(136) a. who loves who(m)?
b. why does she love who(m)?

(139) a. who(m) does who love?

b. who(m) does she love why?
when a sentence contains two wh-expressions, the
superior one (i.e., the one higher in the
structure) moves overtly
(i) Antecedent-Contained Deletion (ACD)
(ii) Weak Crossover (WCO)
(iii) lack of Q-scope ambiguity in wh-movement
cases
(iv) Superiority effects in multiple wh-questions
123
Overt vs Covert Movement
(136) a. who loves who(m)?
b. why does she love who(m)?

(139) a. who(m) does who love?

b. who(m) does she love why?
this follows from the ECP if wh-in-situ must
undergo LF-movt by adjoining to the
wh-constituent that went to SpecCP overtly
(i) Antecedent-Contained Deletion (ACD)
(ii) Weak Crossover (WCO)
(iii) lack of Q-scope ambiguity in wh-movement
cases
(iv) Superiority effects in multiple wh-questions
124
Overt vs Covert Movement
ti is not properly govd
(136) a. who loves who(m)?
b. why does she love who(m)?

(139) a. who(m) does who love?

b. who(m) does she love why?
(140) CP Spec whoi Spec whomjjj C' IP
ti love tj
the subject-wh does not c-command its trace at
LF, and
the index i of the subject-wh does not
percolate to SpecCP
(i) Antecedent-Contained Deletion (ACD)
(ii) Weak Crossover (WCO)
(iii) lack of Q-scope ambiguity in wh-movement
cases
(iv) Superiority effects in multiple wh-questions